Recoloring of visual multimedia using matlab to aid color vision deficient individuals

Abstract

Color Vision Deficiency (CVD) has a widespread impact and has affected more than 300 million people across the world. CVD has three major types as Monochromacy, Dichromacy and Anomalous Trichromacy. We design and develop a website with re-coloring algorithms incorporated for anomalous trichromats and dichromats. Unlike an image recoloring schemes, recoloring a video demands maintaining color consistency in the frames which implies similar color in adjacent frames should be recolored to similar necessary color. Our aim is to show the Color Vision Deficient (CVD) the primary colors distinguishably, which they cannot see in normal lighting conditions. To achieve this goal, we have incorporated recoloring algorithm into the website using Matlab as a back-end language, which will first conduct Ishihara plates test to check if a user is CVD. If yes, the test will then comment on the type of its Color Blindness, based on this user would be recommended to go to the respective module of recoloring. User would be asked to submit video or an image which is to be re-colored. Video/Image would be re-colored on server side, the recolored output would be displayed on a screen with an option to download result. The RGB to LMS algorithm is used to recolor those pixels in images and videos which are unperceivable to users. After recoloring they will be able to distinguish the colors and it will be more recognizable. Finally, there would be a feedback/suggestions which users are asked to fill-out, to understand the performance of algorithm and quality of re-colored image. Experimental results prove that with the help of our website almost all CVD irrespective of their types would be helped in their vision to experience multimedia. Thus our website would aid millions of CVD’s to experience colors in multimedia which they are unable to see distinguishably.

Keywords

Color Vision Deficient (CVD)colour-blindness (CB)color blind unperceivable (CBU)color blind perceivable (CBP)monochromacy dichromacy anomalous trichromacy

1. Introduction

Color-blind deficiency, is found more in men as compared to women. Statistically it’s approximately 5% of the population (i.e. 8% of males and 0.5% of females) or five out of every hundred users who are exposed to computer usage in their daily life. The prominent CVD is a red-green sensitivity or called Protanomaly according to Huang et al. [1]. People with CVD find it difficult to distinguish the certain color combinations impacting their ability to perform tasks related to color and visualization. While a lot of research and pragmatic experimentation has been done in medical and genetics, a perennial solution yet to be found. In fact, all the results have been ephemeral. Therefore, research in the area of image processing to help color blind people distinguish and perceive objects has engendered. This deficiency has been fostered by deformity of rod and cone cells that are situated within the retina as stated by Carlson et al. [3]. Cone cells help to distinguish the color and rod cell regulates the light intensity. Cone cells are of three types they are the red, blue, green cone cell. Color blind deficiency is the disability that can hardly be altered or cured. In most cases, it is hereditary or genetic and the condition stays the same throughout. It doesn’t get better or worse. As technology bloomed and all of a sudden massive demand for multimedia became a necessity in one’s life but CVD miss out on important details in images and videos due to the deficiency in their retina.

They have difficulties in viewing traffic lights, digital images and videos. It happens because of genes which are inherited through the color blind parents. Genes are responsible for it. Discrimination against the color blind deficient individuals is evident at workplaces, even if they carry out the work properly.

Color blind deficiency individuals are more in numbers and their difficulties just cannot be ignored. Each person’s severity of impairment may differ depending on the extent of the shift in the wavelength of color distinction. More than 20 nm shift in spectral response of a cone is essentially equivalent to complete blindness that is received by that specific cone cell. Retina produces the impulse that will be received by the human brain in the occipital lobe and then it will be represented as a color. Carlson et al. [3] believed that a universal condition of the eye can be given by Ishihara test. In our website we have incorporated Ishihara test, which will give the color blindness type as well. The Ishihara test is one of the methods to determine whether the person has CVD or not. However, it’s still unable to categorize a color deficient individual with accurate detail and perfection since its inception in 1917. The Ishihara tests obtainable today justifies whether people have CVD. For any CVD individual to see the recolored image, it is first important to check which primary colors are not easily detected by him/her under normal lighting conditions. Hence the following are various types of color vision deficiencies:

1.
Monochromacy: This vision have either no cones or one cone present in the retina and can see no color. Their vision consists of different shades of grey color.
2.
Dichromacy: Two different cone types are present, the one is missing completely

(a)
Protanopia: CVD cannot see Red color completely (Missing L-cone)
(b)
Deuteranopia: CVD cannot see Green color completely (Missing M-cone)
(c)
Tritanopia CVD cannot see Blue color completely (Missing S-cone).

3.
Anomalous trichromacy:

(a)
Protanomaly: CVD cannot distinguish between Red-Green (Malfunctioning in L/M cone)
(b)
Deuteranomaly: CVD cannot distinguish between Green-Blue (Malfunctioning in M/S cone)
(c)
Tritanomaly: CVD cannot distinguish between Red-Blue (Malfunctioning in L/S cone).

Thus due to an increasing number of CVD individuals daily and no cure being developed so far for color blindness, it is need of humanity to develop some sort of solution which can help the CVD’s to perceive multimedia of all the three primary colors distinguishably even under normal lighting conditions.

The remainder of this paper discusses about the layout of the research work in sections. Section 2 discusses about the background. Section 3 discusses about the proposed methodology, Material and Method are discussed in Section 4, and Section 5 is about numerical analysis. Section 6 is about experimental results and Section 7 highlights about the conclusion of our research work.
2. Background

A number of studies have been reported in literature which is focused on colors which plays a vital part in all sorts of activities. Images are often analyzed by its meta-data like size, type, form etc. Identifying the change in the intensity between two different regions in an image can be done by the edges, which leads to get more details about the object, its size and shape. Dody et al. [4] have discussed about few applications of edge detection including surveillance, text and shape detection. The human eye accommodates two forms of image receptors: namely rods and cones clearly stated by Navada et al. [5]. The rod cells are responsive to dim light levels which do not distinguish between colors. Brighter light levels are responsive because of cones, which allow us to see different colors. Existence of three types of cones make color perception possible: long, medium and short wavelength cones. They each correlate to a specific light wavelengths that epitomize the three basic colors that are red (long wavelength), green (medium) and blue (short) as shown in Fig. 1. Vision that utilizes three receptors for color perception is called as normal vision.

Figure 1.

Wavelength of the cone system depicting normal vision [5].

Color Vision Deficiency can be identified through an objective based eye examination called as Ishihara plates. Liu et al. [12] indicated, the patient sees a series of specially designed images (plates) composed of different close frequency color dots, called pseudo isochromatic plates. Pseudo isochromatic plate testing determines if a color vision deficiency exists and also the form of deficiency.

CVD’s find difficulty in identifying the different colors in traffic signals, fruits, vegetables [11]. They used certain functions to analyze and recolor the entire websites for anomalous trichromats; it also ensures that it maintains the ratio between the color differences; perceived by trichromats and anomalous trichromats. It is based on the genetic algorithm [8].

As far as we know, there are only a handful of works, some of them by Huang et al. [17], Jeong et al. [18], regarding re-coloring of videos for color deficient individuals. In [21], Machado et al. showcase a re-coloring algorithm for enhancing color contrasts for dichromats using GPU. In [1], Huang and Chiu focused on TCC based video reproduction for dichromats specifically. However, they did not guarantee the similar color appearance in different frames to be remapped to the new color uniformly as the temporal relationship between frames was not taken into accountability.

Several research works have been done out to mitigate the problem of unperceivable colors for CVD so they can perceive visual objects with different combinations of colors and color differences. A mapping function can be adapted to change the unperceivable colors by CVD to colors perceived with different colors so they can distinguish it. Recoloring algorithms have two problems first they have huge computational cost as most of the recoloring algorithms use an optimization process to determine the color mapping function and the other flaw is maintaining of naturalness. It arises because of the inconsistency of colors while recoloring of different frames of videos. The various systems that already exist are the Color Correction System invented by A. Thomas who designed contact lenses and glasses for corrective vision. The wavelength of each color that goes into the eyes is changed by the filters, effectively making one to see more colors. The ColorAdd System developed by Miguel Neiva allows CVD for navigation, reference and organization. The EyePlot is operable on Macintosh and Windows Operating Systems.

We are focusing on all types of color blindness except for Monochromacy as they perceive only grey levels, recoloring or wavelength adjustment will not significantly help the CVD’s as they will again only experience grey levels of recolored multimedia.

We conducted the clinical analysis on 300 participants under the guidance of Municipal Corporation of Greater Mumbai’s Department of Ophthalmology, Topiwala National Medical College and B.Y.L. Nair Hospital. The Ethical Review committee performed the ethical review of research and gave a protocol number. “Enabling feature distinction and preserving naturalness in visual media using the combined video processing approach for color deficient individuals Version 1 dated 27 ${}^{\text{th}}$ February 2018”. The Committee also verified the recoloring algorithm incorporated into the website with actual CVD patients testing on the website which marks it as one of the most important contributions to the world with no expectations of profit from it.

Table 1

Types of color blindness and their amount of rareness

Sr. no.	Category	Deficiency name	Defect	Rareness
1.1	Anomalous trichromacy	Protanomaly	L-cone defect	1.3%
1.2	Anomalous trichromacy	Deuteranomaly	M-cone defect	4.9%
1.3	Anomalous trichromacy	Tritanomaly	S-cone defect	0.01%
2.1	Dichromacy	Protanopia	L-cone absent	1%
2.2	Dichromacy	Deuteranopia	M-cone absent	1.1%
2.3	Dichromacy	Tritanopia	S-cone absent	0.02%
3	Monochromacy	Monochromacy	No cone functioning	Very rare

3. Proposed methodology

As discussed in abstract, our recoloring algorithm is incorporated into the website whose functionality is written in Matlab programming language which will check whether the user is CVD or not. If yes then the user is asked to submit the video or image. The advantage of this website is that the recolored image could be processed within seconds and the output would be displayed on a screen with the download option. The limitation of website is that the video is completely broken into frames and each frame is then recolored as per the respective type of color-vision deficiency and then the recolored frames would be then clustered back to form a recolored video, so this complete process requires some decent amount of time. Hence, for processing of videos the user is asked to submit video and check after sometime approximately to get back the recolored video in downloadable format. Timings are discussed in the result section. The following is the block diagram for video/image recoloring.

Figure 2.

Block diagram of recoloring process of our system for both images and videos.

Figure 3.

Flow chart for recoloration of a video.

Now to understand each and every algorithm as per type of color vision deficiency, we must have to take a look at the following table, to understand types of color vision deficiencies and the amount of rareness.

The proposed methodology consists of a method to modify an image/video in order for it to be more perceivable for those suffering from (Dichromacy and Anomalous Trichromacy).

The proposed methodology consists of the algorithm for Images:

(a)

Accept the input image and store it in browser cache

(b)

Extract RGB co-ordinates

(c)

Multiply with respective co-ordinates

(d)

Generate output of each type of deficiency; for the given input image

(e)

Display the image’s output

The proposed methodology consists of the algorithm for Video Recoloring:

(a)

Accept the input video, and store it in a browser cache.

(b)

Extract Audio.

(c)

Break video into shots.

(d)

Convert shot to frames.

(e)

Extraction of frames and loading.

(f)

Differentiation of CBP and CBU colors.

(g)

Re-mapping of the color.

(h)

Recombining of the frames to shots.

We have used File System to store original and processed multimedia files like images and videos to demonstrate an image and its output. An original frame is processed and recolored three times with names as the simulated image, the recolored image and the simulated recolored image. The simulated image means what the color vision deficient perceives from the original image before recoloring it. The recolored image means what the color vision deficient is expected to see and perceive all 3 primary colors. The simulated recolored image is the image, color vision deficient will perceive from the recolored image. The file system clearly defines the input and output of multimedia files. We are using Matlab libraries to process the files. All these images are used to populate the website data for recoloring the multimedia. In nutshell, we are not using any traditional database like MySQL or Oracle.

4. Material and method

This research is carried out by understanding the difficulties faced by CVD individuals. An enormous amount of information is passed through television, cell phones and the internet. This makes the necessity of modifying video and image for CVD users. Therefore, using image processing, the digital image can be enhanced and modified to get more sharpening, removal of noise, rotation, scaling. We apply RGB to LMS re-coloring algorithm based on the threshold value of the pixel to change the color of those pixels that are unperceivable to color deficient individuals.

4.1 Steps of video recoloring

Detection of shots

Each scene is made of several shots. A shot is the continuous footage or sequence between two edits or cuts of a camera roll. Frames are nothing but still images achieved by the shots as in [1], shot detection is performed and each frame of the shot is processed as a single image, after completion of all frames of a shot, the frames of the next shot are processed.

Extraction of frames and loading

A shot consists of several frames which are been detected. Each frame is loaded one at a time and processed individually and re-coloring algorithm is applied on each of them. A color appearing in multiple frames should be remapped to the same color in all frames should be done by the recoloring algorithm.

Differentiating CBP and CBU colors

Colors that are not perceivable to color deficient individuals have to be remapped from that particular frame. We need to distinguish the color blind perceivable colors from the color blind unperceivable colors [6]. A simulated image is created for distinguishing purpose, this is the image how it is seen by the colorblind.

The following procedure is followed:

4.1.1 Pre-processing

Trichromatic vision is if all three cones $L$ , $M$ , and $S$ are present but when it comes to dichromacy, due to the absence of one type of cone, the 3D color space is limited to a plane.

Color plane in LMS Space is defined as:

$\displaystyle\alpha L+\beta M+\gamma S=0$ (1)

Where $\alpha$ , $\beta$ , and $\gamma$ are unknown parameters of a plane in the LMS space.

As per Eq. (1).

For example, in deuteranopia the $L$ and $S$ are unchanged, but new $M$ value is calculated as a function of $L$ and $M$ :

$\displaystyle MD=-(\alpha L+\gamma S)/\beta$ (2)

The following steps are applied to each image pixel represented by RGB color coordinates:

Gamma correction: Gamma is the relationship between RGB intensity values of an image. The relationship between input and output is that output is proportional to the input raised to the power of gamma.

$\displaystyle[R,G,B]=[R/255,G/255,B/255]^{\wedge}2.2$

Using LMS Color Space for recolorization

$L$ , $M$ , $S$ cones are responsible for color in the human retina. So first we need to calculate the LMS values of the RGB image using conversion matrix.

RGB corresponds to the phosphor levels on a cathode ray tube screen to match the colors used to define digital images.

Transformation of RGB to XYZ to LMS:

$\displaystyle\left[\begin{array}[]{c}L\\ M\\ S\\ \end{array}\right]=\left[\begin{array}[]{lll}{17.8824}&{43.5161}&{4.11935}\\ {3.45565}&{27.1554}&{3.86714}\\ {0.0299566}&{0.184309}&{1.46709}\\ \end{array}\right]\times\left[\begin{array}[]{c}R\\ G\\ B\\ \end{array}\right]$

Transformation of 3D LMS space to dichromats 2D spaces solving the plane equation for deuteranopes:

$\displaystyle\left[\begin{array}[]{c}Ld\\ Md\\ Sd\\ \end{array}\right]=\left[\begin{array}[]{lll}{1}&0&0\\ {0.494207}&0&{1.24827}\\ 0&0&{1}\\ \end{array}\right]\times\left[\begin{array}[]{c}L\\ M\\ S\\ \end{array}\right]$

Inverse transform LiMiSi to XYZ to RGB, $i=\{P,D,T\}$ :

$\displaystyle\left[\begin{array}[]{c}R\\ G\\ B\\ \end{array}\right]=\left[\begin{array}[]{ll}{0.080944}&{-0.130504}\\ {0.116721}\\ {-0.0102485}&{0.0540194}\\ {-0.113615}\\ {-0.000365294}&{-0.00412163}\\ {0.693513}\\ \end{array}\right]\times\left[\begin{array}[]{c}Li\\ Mi\\ Si\\ \end{array}\right]$

Inverse gamma correction:

$\displaystyle[R,G,B]=255\times([R,G,B]^{\wedge}1/2.2)$

Similarly conversion matrix for protanopes is:

$\displaystyle\left[\begin{array}[]{c}Lp\\ Mp\\ Sp\\ \end{array}\right]=\left[\begin{array}[]{lll}0&{2.02344}&-2.52581\\ 0&{1}&0\\ 0&0&{1}\\ \end{array}\right]\times\left[\begin{array}[]{c}L\\ M\\ S\\ \end{array}\right]$

Similarly conversion matrix for tritanopes is

$\displaystyle\left[\begin{array}[]{c}Lt\\ Mt\\ St\\ \end{array}\right]=\left[\begin{array}[]{lll}{1}&0&0\\ 0&1&0\\ {-0.395913}&0&{-0.801109}\\ \end{array}\right]\times\left[\begin{array}[]{c}L\\ M\\ S\\ \end{array}\right]$

The complete algorithm of tritanopia is explained in the Section 6.3.

In step [4], detection of the unperceivable colors is done by first applying the red mask on the pixels. As the colors get separated into subsets of perceivable and unperceivable colors. The $D_{t}$ value is calculated. It is the difference of the pixel distance between the different frames i.e. adjacent frames. It is applied on the same co-ordinate values to improvise the contrast. If this $D_{t}$ distance is greater than a particular threshold value, then a spatial color mapping strategy is applied otherwise a temporal color mapping strategy is applied. The pixel is rotated by the certain angle which will remap it to color blind perceivable color is the spatial color mapping. Global and Local rotations are performed. If it is less than a threshold value then a temporal color mapping strategy is applied where a similarity searching algorithm is used to find the color which is most similar to the current pixel value in the previous frame. Application of linear prediction is also done.

4.1.2 Colors remapping

From the above steps we know which pixels are unperceivable we have to remap them [1]. For this we use two strategies as explained in step no 4.

Which strategy is to be used depends on the temporal color difference ( $D_{t}$ ). The temporal color difference is the distance of the pixel value between the pixels to be converted for color mapping of the pixel in the previous frame at the same image coordinate. If the temporal color difference is greater than a threshold value then we use spatial color mapping strategy else we use temporal color mapping strategy

Figure 4.

Global rotation of color component X.

4.1.2.1 Spatial Color Mapping Strategy (CMS ${}_{S}$ )

In CMS’s each CBU color X is recolored to X’ with the angle ${\Delta}\theta$ for color rotation. Here ${\Delta}\theta$ is the sum of two angles and is defined as,

$\displaystyle\Delta\theta=\theta^{G}+\theta^{L}$ (3)

Where $\theta^{G}$ is the global rotation angle and $\theta^{L}$ is the local rotation angle.

The global rotation angle is for discrimination of the color blind perceivable colors from the color blind unperceivable colors. Hence, the global rotation angle is directly proportional to the difference between the original pixel value and value for a simulated pixel.

The local contrast among CBU colors needs to be enhanced so the local rotation angle ( $\theta^{L}$ ) is applied. The perceived brightness of adjacent regions gets affected by the contrast at their borders. Hence to magnify the perceived contrast of CBU region local rotation is performed.

4.1.2.2 The Temporal Color Mapping Strategy (CMS ${}_{T}$ )

This method is adopted when there is not large distance between the original pixel color and simulated pixel color. It is done to reduce computational cost and at the same time maintain temporal color consistency. In this method, we apply color prediction process to find the pixel that has the most similar color in the previous frame. Then linear color prediction (LCP) is used to find the remapped color.

The process of LCP is as follows, let $X_{t}$ be an input pixel, $X_{t-1}^{*}$ be the most similar color in the last frame and $X_{t-1}$ is pixel at the same position in previous frame.

Then remapped color X’ is defined as:

$\displaystyle X^{\prime}=\left[\begin{array}[]{c}R^{\prime}\\ G^{\prime}\\ B^{\prime}\\ \end{array}\right]$ (4)

To increase computational efficiency even further, we may take a second threshold such that if the linear difference is below this threshold value then directly map $X_{t-1}^{*}$ to $X_{t}$ without performing linear prediction

4.1.3 Recombining frames

Finally, after all the frames of a shot are transformed, these frames are recombined to form the particular shot. This shot will then be played. In case of image recoloring, the image is first loaded, then CBU and CBP colors are distinguished just as they are done in video recoloring. Next remapping of CBU is done using spatial color mapping strategy. The image produced after this stage is the transformed image.

5. Numerical analysis

1.
Once the shot detection is done the ith shot $S_{i}$ of the video is considered as a set of different frames as follows: $S_{i}=\{f_{m},f_{m+1},\ldots,f_{n}\}$ where the indices of the first and last frames of $S_{i}$ are denoted by $m$ and $n$ .
2.
As the shots are achieved from the video then extraction of colors from each shot needs to be done. Shot volume in reality remains is too large so assembling all the pixel values for clustering shot key colors needs to be also considered. Elimination of similar pixel values between frames which are adjacent can be done in advance. For pruning of the repetitive colors the motion estimation approach [22] is applied. As frames in the same shot are quite similar most image blocks in the tth frame $f_{t}$ can be represented by those in the previous frames $f_{t-1}$ . So only few of memory storage is required to store the residual image blocks after deleting repetitive image blocks.
3.
The number of colors to be quantized in $f_{t}$ is related to the ratio of residual image blocks. Let the first frame $f_{m}$ of $S_{i}$ be quantized into 256 colors. The number of colors to be quantized in $f_{t}$ is computed as follows:

$\displaystyle Qf_{t}=256\times\frac{\text{\# of residual blocks in }f_{t}}{% \text{\# of blocks in }f_{t}}$

The total no $Q_{i}$ of quantized colors in $S_{i}$ can be computed as follows:

$\displaystyle Qi=256+\sum^{n}_{t=m+1}Qf_{t}$
4.
We compute the inter color change rate of frames, color reduction score and contrast reduction score. Contrast reduction score is the ratio of the mean contrasts of the original and remapping versions to describe quantitatively the difference in contrast. Perceptual distances are considered between the pairs of a point and its neighbor point at one pixel apart to compute color contrast. If the color contrast after remapping are close the original one, it implies the remapping method can maintain the color contrast. The distinguishable colors mean the colors perceived by viewers. The color reduction score is the ratio of the mean number of distinguishable colors of the original and the remapping versions of videos. The ICCR score is the inter color change rate of frame is used to evaluate the color changes between the frames. Let $D_{t}$ be the number of distinguishable colors of the frame $F_{t}$ and $F_{t+1}$ . ICCR is defined as:

$\displaystyle\textit{ICCR}=\frac{D_{\textit{gone}}+D_{\textit{come}}}{D_{t}% \bigcup{D_{t+1}}}\times 100\%$

where $D_{\textit{gone}}=D_{\textit{t}}-D_{{t}}\cap D_{t+1}$ and $D_{\textit{come}}=D_{t+1}-D_{t}\cap D_{t+1}$ .

The mean of colors contrast is the average of the color contrast of all frames of each video. While the contrast reduction score is closer to one, the reproduced video is more similar to the original video.
5.
In the implementation of the proposed algorithm, the color accessibility in the image-level is evaluated. The Naturalness indicates how much distortion occurs through the recoloring process is calculated by,

$\displaystyle\textit{NAT}=\frac{1}{n(0_{1}U0_{2})}\times\left(\sum_{x^{2}\in Ft% }||X^{\circ}-\hat{X}^{\circ}||^{2}\right)$

Where $n(A)$ is the number of elements in the set $A$ . The smaller NAT indicates the better preservation of color information. The Local Detail and the Color Reduction Score are the measures of the color contrast and its preservation. The Naturalness assessment term is in the scale of 0–10.

6. Results

6.1 Protanopia and protanomaly

As these types of CVDs cannot see the red color completely or confuse between the red and the green color. It is very difficult for them to do daily activities which involves the red color in it. The best example being traffic signal or any danger signs; which are usually in the red color. The following algorithm proposed for the Protanopic CVD which will help in the way that CVD will be able to get a basic difference between the red shades it perceives and after recoloring will experience those shades in a better way to understand the overall content of the image. The efficiency of the algorithm is increased by considering the following factors:

1.
Increasing the difference between the color blind unperceptive and color blind perceptive colors, improves the color perceptibility of the image.
2.
A real-time operation is necessary.
3.
Temporal color consistency be preserved.
4.
Enhancement of local contrast in the region of the color blind unperceptive color needs to be done.

Algorithm which is designed is much more efficient than the existing techniques of re-coloring with the advantage of recoloring a video of both types be it 30 FPS (Frames per second) or 60 FPS. The Algorithm is further compared on the basis of experimental results with other existing techniques of recoloring in the table based on processing time for generating output recolored the image as the primary parameter. The following is the block diagram for algorithm.

Figure 5.
(a) An original brain image, (b) Simulated image (Original image as seen by protanopia), (c) Recolored image, (d) Simulated recolored image as seen by protanopia.

Figure 6.
(a) An original gaugin image, (b) The simulated image, (c) The recolored image, (d) The simulated recolored.

Figure 7.
Carom video results (a) An original image, (b) The simulated image sequence for protanopia, (c) The recoloring method, (d) The simulated recolored frames.

The results of above algorithm for Protanopia in image and video are explained in the detail manner with respect to images and videos with various important coefficients such as Mean, Naturalness and Naturalness assessment as well as quality factors are also given such as Contrast Reduction Score, Color Reduction score and ICCR are explained in detail.
6.1.1 Image results for protanopia individuals

The figure given above displays the result of the proposed system for brain image (Fig. 5a). As protanopes are unable to differentiate between red and black, they cannot distinguish between regions in the image (Fig. 5b). To overcome this drawback, we recolored the image (Fig. 5c). After re-coloring the person suffering from protanopia will be able to differentiate between the red and black region (Fig. 5d).

The figure given above displays, the result of the proposed system for a Gaugin image (Fig. 6a). As protanopes are unable to differentiate between brownish red and green in the given plane, they cannot distinguish between regions in the given plane (Fig. 6b). To overcome this drawback, we re-colored the image (Fig. 6c). After re-coloring person suffering from protanopia will be able to differentiate between the regions of the plane (Fig. 5d).

Table 2
Some important coefficients for recolored image

Image	Quantized colors	Contrast reduction score	Number of colors	Color reduction score
Brain	1024	1.25	6518	4.32
Gauguin	1024	1.32	5497	4.58

Table 3

Quality factors and co-efficient for the recolored image

Image	Resolution	Format	Compression type	Processing time (secs)
Brain	235 $\times$ 215	PNG	Highly Compressed	7.3532
Gauguin	370 $\times$ 294	PNG	Highly Compressed	8.2154

6.1.2 Video results for protanopia individuals

As shown in the above Fig. 7b, 17 ${}^{\text{th}}$ , 76 ${}^{\text{th}}$ , 165 ${}^{\text{th}}$ , 234 ${}^{\text{th}}$ frames the red color queen coin cannot be seen by the protanopes. As protanope is unable to see red color queen in the frame no 17 ${}^{\text{th}}$ and frame no 76 ${}^{\text{th}}$ but the problem arises when a black color coin also moves near the red color queen. Protanope cannot distinguish between the coins. As the coin appears to s/he as black and protanopes is getting confused between the coins. The recoloring is performed in Fig. 7c, the contrast of the red color queen coin is increased as compared to black color coin by recoloring algorithm. Finally after recoloring protanope can differentiate between the two different colors coins.

Table 4
Coefficients for quality of recolored video

Frame number	Resolution	Format	FPS (Frames per second)	Total frames
17	640 $\times$ 360	AVI	30	323
76	640 $\times$ 360	AVI	30	323
165	640 $\times$ 360	AVI	30	323
234	640 $\times$ 360	AVI	30	323

Table 5

Complex coefficients calculation for performance assessment

Frame number	ICCR	Mean	NAT	Naturalness assessment
17	16.254	438.4	25.098	3.5492
76	16.27	447.3	26.124	3.6521
165	16.34	454.7	24.986	3.5478
234	16.47	434.4	24.684	3.5684

Table 6

Complex co-efficient for performance parameters

Frame number	Quantized colors	Contrast reduction score	Number of colors	Color reduction score
17	1024	1.22	5467	4.26
76	1024	1.28	6457	4.43
165	1024	1.27	4879	4.35
234	1024	1.38	4238	4.79

The results for Protanomaly are given in the image as well in video are as follows.

6.1.3 Image results for protanomaly individuals

Figure 8.

(a) An original image, (b) The simulated image, (c) The simulated recolored image, (d) The recolored image.

Figure 9.

(a) An original image, (b) The simulated image, (c) The recolored, (d) The simulated recolored image.

Figure 10.

(a) An original image, (b) Simulated image, (c) Recolored, (d) Simulated recolored.

Figure 11.

(a) An original image (b) Simulated image (c) Recolored (d) Simulated recolored.

6.1.4 Discussions about the results and observations of protanomaly

Table 7
Results of a recoloring images for protanomaly

Image	Resolution	Format	Compression type	Processing time (secs)
Man	180 $\times$ 280	JPG	Highly compressed	7.23
Stats Graph	673 $\times$ 303	PNG	Compressed	8.45
Bar-Graph	380 $\times$ 240	JPG	Highly compressed	8.86
Match	640 $\times$ 480	JPEG	Compressed	7.25

Table 8

Coefficients for recolored image

Image	ICCR	Mean	NAT	Naturalness assessment
Man	21.254	489.336	19.3549	3.8952
Stats Graph	24.578	478.964	18.6451	4.3210
Bar-Graph	18.457	492.362	14.3365	4.2156
Match	54.356	487.694	17.5649	3.5201

Table 9

Quality factors and co-efficient for recolored image

Image	Quantized colors	Contrast reduction score	Number of colors	Color reduction score
Man	1024	1.27	6124	4.36
Stats Graph	1024	1.25	5487	4.37
Bar-Graph	1024	1.26	6894	4.46
Match	1024	1.42	7849	4.87

6.1.5 Video results of protanomaly

Figure 12.

The basketball video results. (a) An original image, (b) The simulated image sequence for protanopia, (c) The recoloring method, (d) The simulated recolored frames.

Table 10

Co-efficient for quality of recolored video

Frame number	Resolution	Format	FPS (Frames per second)	Total frames
10	540 $\times$ 480	AVI	60	352
60	540 $\times$ 480	AVI	60	352
159	540 $\times$ 480	AVI	60	352
294	540 $\times$ 480	AVI	60	352
350	540 $\times$ 480	AVI	60	352

Table 11

Complex coefficients calculation for performance assessment

Frame number	ICCR	Mean	NAT	Naturalness assessment
10	19.35	447.2	25.234	1.2645
60	19.58	437.3	24.548	1.3524
159	19.48	452.6	25.297	1.2015
294	19.42	442.7	23.984	1.2035
350	19.52	450.6	25.892	1.2044

Table 12

Complex coefficients for performance parameters

Frame number	Quantized colors	Contrast reduction score	Number of colors	Color reduction score
10	1024	1.48	5846	4.82
60	1024	1.28	6846	4.26
159	1024	1.34	9817	4.44
294	1024	1.33	4567	4.38
350	1024	1.54	9889	4.48

6.2 The deuteranopia and deuteranomaly

As these types of CVD’s cannot perceive green completely or partially confuse between green and blue color, they face problems in daily activities which involves green wavelength like choosing between green leafy vegetables and fruits, etc. not experiencing green color in multimedia or even in live events like matches. Now firstly, focusing on Deuteranomaly where the CVD confuse to see between the green color and the blue/red color completely under normal lighting conditions. The following is the algorithm for Deuteranomaly recoloring of video/image.

Algorithm:

1.
First the color blind user accesses our website.
2.
We ask the user if s/he suffering from deuteranopia.
3.
Then we ask the user to browse the image or video that he/she wants to process
4.
We take a copy of this image as the input for our algorithm & keep the original image as it is.
5.
Then we use our algorithm to process the image and if the user take a video as input then process the generated frames.
6.
We now get the recolored image or frames suitable for the color blind user as output.
7.
We give this image as output to the user and ask the user if he/she wants to save the image or video frames.

6.2.1 Image results for deuteranopia

Figure 13.

(a) An original image of Test-Cube, (b) The simulated image (as seen by deuteranopia), (c) The recolored image, (d) The simulated recolored image as seen by deuteranopia.

The figure given above displays the result of the proposed system for Test-Cube original image (Fig. 13a). As deuteranopia cannot perceive the green color (Fig. 13b) is the simulated image created by our algorithm. To overcome this drawback, we re-colored the image (Fig. 13c). After re-coloring the person suffering from deuteranopia will be able to view green as different color recolored to brown color. This type of CVD will be able to differentiate between the planes of the cube (Fig. 13d). It is the beauty of our algorithm we are able to recolor the image as per the simulated image which is been perceived by the deuteranopia.

Figure 14.

(a) An original image of Test-Spheres, (b) The simulated image (seen by deuteranopia), (c) The recolored image, (d) The simulated recolored image as seen by deuteranopia.

The figure given above displays the result of the proposed system for the test-spheres original image (Fig. 14a). As deuteranopia cannot perceive green color balls (Fig. 14b) is the simulated image created by our algorithm. To overcome the drawback, we recolored the image (Fig. 13c). After re-coloring the person suffering from deuteranopia will be able to view green as different color recolored to brown color. This type of CVD will be able to differentiate between the different colors of balls (Fig. 14d). The beauty of our algorithm is we are able to recolor the image as per the simulated image which is perceived by the deuteranopia.

Table 13

Results of recoloring image for deuteranopia

Image namImage	Resolution	Format	Compression type	Processing time (secs)
Test cube	700 $\times$ 500	JPG	Highly Compressed	9.43
Test spheres	215 $\times$ 144	JPG	Highly Compressed	8.89
Test flag	163 $\times$ 154	JPG	Highly Compressed	8.21

Table 14

The various important coefficients for recolored image

Image	ICCR	Mean	NAT	Naturalness assessment
Test cube	17.2548	389.457	17.2546	4.6548
Test spheres	37.5489	387.346	16.2549	4.5699
Test flag	45.3545	388.124	17.5546	4.0254

Table 15

Quality factors and coefficients for recolored image

Image	Quantized colors	Contrast reduction score	Number of colors	Color reduction score
Test cube	1024	1.34	2545	5.43
Test spheres	1024	1.33	3595	5.21
Test flag	1024	1.43	2894	5.79

Figure 15.

(a) An original image of Test-Flag, (b) The simulated image (seen by deuteranopia), (c) The recolored image, (d) The simulated recolored image as seen by deuteranopia.

The figure given above displays result of the proposed system for the test-flag original image (Fig. 15a). As deuteranopia cannot perceive green color region (Fig. 15b) is the simulated image created by our algorithm which is perceived by deuteranopia. To overcome this drawback, we re-colored the image (Fig. 15c). After re-coloring the person suffering from deuteranopia will be able to view green, our algorithm recolors to dark brown color. This type of CVD will be able to differentiate between the different colors of regions in the flag (Fig. 15d).

6.2.2 Results of video for deuteranopia

Figure 16.

(a) original video frames for deuteranopia, (b) frames seen by CVD, (c) Recolored frames seen by CVD.

Table 16

The description of input video for deuteranopia

Frame number	Resolution	Format	FPS (Frames per second)	Total frames
13	540 $\times$ 480	MKV	60	314
47	540 $\times$ 480	MKV	60	314
87	540 $\times$ 480	MKV	60	314
156	540 $\times$ 480	MKV	60	314
198	540 $\times$ 480	MKV	60	314

Table 17

Coefficients for quality of recolored video for deuteranopia

Frame number	ICCR	Mean	NAT	Naturalness assessment
13	25.2145	378.364	19.784	2.3251
47	25.3154	369.254	19.254	2.4814
87	25.3648	376.497	20.002	2.2548
156	25.1569	386.361	19.765	2.8477
198	25.3156	387.467	20.315	2.2514

Table 18

Complex coefficients calculations for performance assessment

Frame number	Quantized colors	Contrast reduction score	Number of colors	Color reduction score
13	1024	1.44	5847	5.78
47	1024	1.25	7698	5.34
87	1024	1.28	17580	5.43
156	1024	1.39	3645	5.67
198	1024	1.24	6815	5.24

6.2.3 Results of videos for Deuteranomaly

Figure 17.

Basket Ball video results. (a) An original image, (b) The simulated image sequence for Deuteranomaly, (c) The recoloring method, (d) The simulated recolored frames.

Hence we can see that the recoloring processing time for an image is very less like within few seconds, but for videos the process of breaking videos into the frames and then reoloring each frame is taking a lot of CPU processing type and then taking all the reolored frames and clustering them to form the original video back takes another decent chunk of time. So only in case of videos the user is asked to submit videos and then collect the recolored video after sometime.

Table 19

Description of Input Video for Deuteranomaly

Frame number	Resolution	Format	FPS (Frames per second)	Total frames
10	640 $\times$ 480	AVI	30	98
20	640 $\times$ 480	AVI	30	98
30	640 $\times$ 480	AVI	30	98
50	640 $\times$ 480	AVI	30	98
70	640 $\times$ 480	AVI	30	98
77	640 $\times$ 480	AVI	30	98

Table 20

Coefficients for quality of recolored video for Deuteranomaly

Frame number	ICCR	Mean	NAT	Naturalness assessment
10	29.212	362.35	22.7851	2.96
20	29.324	354.65	22.5469	2.87
30	29.102	352.21	21.8974	2.54
50	29.60	360.01	23.0215	2.64
70	29.31	357.25	21.5698	2.36
77	29.30	354.57	23.25164	2.48

Table 21

Complex coefficients calculation for performance assessment

Frame number	Quantized colors	Contrast reduction score	Number of colors	Color reduction score
10	2048	1.44	4320	5.76
20	2048	1.33	76918	5.46
30	2048	1.32	5479	5.32
50	2048	1.22	36187	5.22
70	2048	1.37	3798	5.86
77	2048	1.18	4851	5.10

6.3 The tritanopia and tritanomaly

As these types of CVDs cannot perceive completely the Blue color or confuse between Blue and other two primary colors which are Red and Green, they face a lot of difficulties in experiencing visual media like sports (sports match between India and Sri Lanka that is teams having jerseys with Blue shades), adventurous TV shows like Man vs Wild or Blades of the River where the CVD won’t be able to perceive blue shades. Though the percentage of Tritanomaly and Tritanopia suffering CVD are very very less, yet they must be helped experiencing visual media in which they can be able to distinguish colors of blue shades and their respective type. Thus our algorithms help the above mentioned types of CVD to perceive some other color in place of blue but distinguish it with other colors and thus experience the contents of visual multimedia in a better way. The following are the steps for Tritanomaly recoloring videos and images.

1.
Convert the video into individual frames
2.
Extract the RGB color space of a frame
3.
Convert the RGB into LMS color space

The transformation of RGB to XYZ to LMS:

$\displaystyle\left[\begin{array}[]{c}L\\ M\\ S\\ \end{array}\right]=\left[\begin{array}[]{lll}{17.8824}&{43.5161}&{4.11935}\\ {3.45565}&{27.1554}&{3.86714}\\ {0.0299566}&{0.184309}&{1.46709}\\ \end{array}\right]\times\left[\begin{array}[]{c}R\\ G\\ B\\ \end{array}\right]$
4.
Transform to colorblind LMS values

$\displaystyle\left[\begin{array}[]{c}L_{t}\\ M_{t}\\ S_{t}\\ \end{array}\right]=\left[\begin{array}[]{lll}{1}&0&0\\ 0&1&0\\ {-0.395913}&0&{0.801109}\\ \end{array}\right]\times\left[\begin{array}[]{c}L\\ M\\ S\\ \end{array}\right]$
5.
Transform the new LMS values back to RGB Values

$\displaystyle\text{newRGB}=\text{lmst}\times\text{inverse (rgb2lms)}$
6.
Calculate the error between original RGB and new RGB
7.
Modify the Error for people with Tritanopia
8.
Add the error to original RGB

$\displaystyle\textit{err2mod}=\left[\begin{array}[]{c}0.7\ \ 1\ \ 0\\ 0.7\ \ 1\ \ 0\\ \ 0.7\ \ 0\ \ 0\\ \end{array}\right]$
9.
Merge the individual frames to finally make the new video.
10.
End

The algorithm designed is better in efficiency and accuracy as compared to existing techniques.

Hereby attached are the image and video results and following them is the table which shows the processing time required for successful recoloring.
6.3.1 Image results

Figure 18.

(a) An original image, (b) The simulated image, (c) The recolored image, (d) The simulated recolored image.

Figure 19.

(a) An original image, (b) The simulated image, (c) The recolored, (d) The simulated recolored image.

Figure 20.

(a) An original image, (b) The simulated image, (c) The recolored image, (d) The simulated recolored image.

Figure 21.

(a) An original image, (b) The simulated image, (c) The recolored image, (d) The simulated recolored image.

Table 22

Results of recoloring an image for tritanopia

Image	Resolution	Format	Compression type	Processing time (secs)
Gaugin	540 $\times$ 380	JPG	Highly Compressed	8.32
Color pencils	740 $\times$ 580	PNG	Compressed	9.54
Color circle	640 $\times$ 480	JPEG	Highly Compressed	8.21
Color smoke	480 $\times$ 320	JPG	Compressed	8.25

Table 23

Various important coefficients for recolored image

Image	ICCR	Mean	NAT	Naturalness assessment
Gaugin	22.454	482.325	12.349	3.3452
Color pencils	24.652	512.254	14.641	4.2310
Color circle	21.467	489.325	12.365	2.7671
Color smoke	22.458	498.254	14.549	3.1091

Table 24

Quality factors and coefficients for recolored image

Image	Quantized colors	Contrast reduction score	Number of colors	Color reduction score
Gaugin	2048	1.17	8478	4.36
Color pencils	2048	1.36	6589	4.37
Color circle	2048	1.40	2154	4.46
Color smoke	2048	1.29	3489	4.87

6.4 Website

We are populating the output in the form of website. The images used in the website are derived using the above algorithms. This enables CVDs to perceive the visual multimedia, except for Monochromacy. This is due to their inability to view any color, they see only grey-levels. Thus our website will contain recoloring algorithms for the following:

1.
Protanopia
2.
Protanomaly
3.
Deuteranomaly
4.
Deuteranopia
5.
Tritanopia
6.
Tritanomaly

The proposed website (http://www.colorblindnp. com) conducts the ishihara plates test to check whether the user is color vision deficient. If positive for any type of deficiency then the user will be directed to the respective type of module as defined by the ishihara plates test. If not, they are asked to submit the feedback rating for the ishihara test which they just conducted.

Figure 22.
The welcome page of website.

Figure 23.
Ishihara test of our website.

Figure 24.
Normal vision result of our ishihara test.

Figure 25.
Protanopic CVD result by Ishihara test.

Post this, on the basis of the assessment of a user’s input/feedback, the respective color blind recolor model is selected automatically and the user is asked to submit the desired video/image for recolor processing.

If the user submits an image; recoloring process starts on the server side and the recolored output image is generated within seconds and displayed on the website. The user is also given an option to download. If the user submits a video then the recoloring process might take some more time depending on the size of the video (as video processing involves breaking of frames, assuming each frame as an image, recoloring each frame and then clustering the recolored frames back to form the recolored video).

The following are screenshots of Ishihara plates test implemented in our website.
7. Conclusion

Thus in this paper, we have discussed the results of recolored images and videos of all the types of color blindness except Monochromacy. The algorithms are implemented in MATLAB programming language; hence the efficiency of getting the results is in quick time as can be inferred from the above results. Image output is still visible in few seconds, but for videos, the execution time is much on a higher side. Also the video recoloring of Tritanomaly/Tritanopia is not included in this paper as they would be included in the next paper of our research work. Also image results for Deuteranomaly and Tritanomaly are not included. We are planning to implement all the above algorithms in Python programming language to improve the output generating capacity of the website. Thus efficiency is expected to increase by 50% for videos. The succeeding paper would be the final results paper which would describe the website and results of algorithms implemented in Python Programming Language.

Footnotes

Acknowledgments

I am thankful to the Ethics Committee for Academic Research Projects (ECARP), the PG Academic Committee of T.N. Medical College and B.Y.L. Nair Hospital. Dr. Renuka Munshi of the Pharmacology Department of T.N. Medical College and B.Y.L. Nair Hospital for her guidance to improvise Algorithm Protocol and Methodology for ICD of Color Blind patients.

References

Huang

Chiu

Chen

. Temporal color consistency-based video reproduction for dichromats. IEEE Transactions on Multimedia. 2011; 13(5), 950–960.

Machado

Oliveira

. Real-time temporal-coherent color contrast enhancement for dichromats. Compute. Graph. Forum. 2010; 29(3), 933–942.

Carlson

. Psychology: The Science of Behaviour (7th Edition). New Jersey, USA: Pearson Education, 2007.

Utama

Latifah

Mengko

Aulia

. Color Blind Test Quantification using RGB Primary Color Cluster. Proc. Int. Conf. on Information Technology Systems and Innovation (ICITSI) Bandung – Bali. 2016; ISBN: 978-1-5090-2449-0.

Navada

Santhosh

Prajwal

Shetty

. An Image Processing Technique for Color Detection and Distinguish Patterns with Similar Color: An aid for Color Blind People. Proceedings of International Conference on Circuits, Communication, Control and Computing, 2014.

Poret

Dony

Gregori

. Image Processing for Color Blindness Correction. IEEE Proc. Int. Conf. Science and Technology for Humanity, (TIC-STH’09). 2009; p. 544. doi: 10.1109/TIC-STH.

Pendhari

Sedamkar

. Multimedia Recoloring Technique for Protanopic CVD. IOSR Journal of Computer Engineering (IOSR-JCE) 2018; e-ISSN: 2278-0661, p-ISSN: 2278-8727, 57–63.

Jeong

Kim

Wang

Jung

. Real-time video recoloring algorithm considering the temporal color consistency for the color – blind. IEEE Transactions on Consumer Electronics. 2012; 58(2).

Kuhn

Oliveira

Fernandez

LAF

. An efficient naturalness-preserving image-recoloring method for dichromats. IEEE Trans. Vis. Comput. Graph. 2008; 14(6), 1747–1754.

10.

Brettel

Vienot

Mollon

. Computerized simulation of color appearance for dichromats. J. Optic. Soc. Amer. A. 1997; 14(10), 2647–2655.

11.

Ichikawa

Tanaka

Kondo

Hiroshima

Ichikawa

Tanabe

Fukami

. Web-page color modification for barrier-free color vision with genetic algorithm. Springer Lecture Notes in Computer Science. 2003; 2724, 2134–2146.

12.

Liu

Wang

Yang

Hua

. Efficient image and video recoloring for colour blindness. Proc. IEEE Int. Conf. Multimedia and Expo (ICME’09). 2009; pp. 906–909.

13.

Orii

Kawano

Maeda

Kouda

. Color Conversion Algorithm for Color Blindness Using Self-Organizing Map. IEEE Proc. Int. Conf on Soft Computing and Intelligent Systems (SCIS – ISIS). 2014; doi: 10.1109/SCIS-ISIS.2014.7044811.

14.

Ruminski

Bajorek

Ruminska

Wtorek

Bujnowski

. Computerized Color Processing for Dichromats. Proceedings of Human – Computer Systems Interaction. AISC 98. Part I. 453–470.

15.

Huang

Tseng

Wang

. Information pre-serving color transformation for protanopia and deuteranopia. IEEE Signal Processing Letter. 2007; 14(10), 711–714.

16.

Hayati

Othman

Sabudin

. A Study of Color Transformation for Color Deficient Individual. IEEE Student Conference on Research and Development (SCOReD). 2013. doi: 10.1109/SCOReD.2013.7002601.

17.

Huang

Lee

Chen

. Shot change detection via local key point matching. IEEE Trans. Multimedia. 2008; 10(6), 1097–1108.

18.

Jeong

Yoon

Kim

. Color Modification For Color-Blind Viewers Using The Dynamic Color Transformation. IEEE Proc Int Conf. on Consumer Electronics. 2012. doi: 10.1109/ICCE.2012.6162036.

19.

Anagnostopoulos

Sekouras

Kalloniatis

Anagnostopoulos

. Intelligent Modification for the daltonization process of digitized paintings. Proc. Int. Conf.on Computer Vision Systems: 2007; Applied Computer Science Group. Bielefeld University, Germany, ISBN 978-3-00-020933-8.

20.

Doliotis

Athitsos

Anagnostopoulos

Tsekouras

. Intelligent Modification of colors in digitized paintings for enhancing the visual perception of color-blind viewers. Proc. Int. Conf. on Artificial Intelligence Applications and Innovations. 2009. pp. 293–301.

21.

Machado

Oliveira

Fernandez

. A physiologically-based model for simulation of color vision deficiency. IEEE Transactions on Visualization and Computer Graphics. 2009; 15(6), 1291–1298. doi: 10.1109/tvcg.2009.113.

22.

Lin

Tai

. Fast full-search block-matching algorithm for motion-compensated video compression. IEEE Trans. Commun. 1997; 45(5), 527–531.

23.

https://nei.nih.gov/health/color_blindness/facts_about.

24.

Pendhari

Dabhoiwala

Mazgaonkar

Tole

. Video and Image Re-Coloring For the Color Vision Deficient. Int. Journal of Global Technology Initiatives. 2015.

25.

Kovalev

. Mining dichromatic colours from videos. Proc. Industrial Conference. Data Mining. 2006; pp. 431–443.

26.

Yang

Wong

Lee

. Quantification and standardized description of color vision deficiency caused by anomalous trichromats – Part I: Simulation and measurements. EURASIP Journal of Image Video Processing. 2008.

27.

Yang

Wong

Lee

. Quantification and standardized description of color vision deficiency caused by anomalous trichromats – Part II: Modelling and color compensation. EURASIP Journal of Image Video Processing. 2008.

28.

Jefferson

Harvey

. An interface to support color blind computer users. In Proc. SIGCHI Conf. Human Factors in Computing Systems (CHI’07), 2007; pp. 1535–1538.

29.

Wakita

Shimamura

. Smartcolor: Disambigous framework for the colorblind. Proc. 7

{}^{\text{th}}

Int. ACM SIGACCESS Conf. Computers and Accessibility (Assets’05), 2005, pp. 158–165.

Recoloring of visual multimedia using matlab to aid color vision deficient individuals

Abstract

Keywords

1. Introduction

4.1 Steps of video recoloring

Detection of shots

Extraction of frames and loading

Differentiating CBP and CBU colors

4.1.1 Pre-processing

4.1.2.1 Spatial Color Mapping Strategy (CMS S )

4.1.2.2 The Temporal Color Mapping Strategy (CMS T )

5. Numerical analysis

6.1 Protanopia and protanomaly

Table 2 Some important coefficients for recolored image

Table 4 Coefficients for quality of recolored video

Table 7 Results of a recoloring images for protanomaly

Footnotes

Acknowledgments

References

4.1.2.1 Spatial Color Mapping Strategy (CMS ${}_{S}$ )

4.1.2.2 The Temporal Color Mapping Strategy (CMS ${}_{T}$ )

Table 2
Some important coefficients for recolored image

Table 4
Coefficients for quality of recolored video

Table 7
Results of a recoloring images for protanomaly