Performance Evaluation of Color Retinal Image Quality Assessment in Asymmetric Channel VQ Coding

Performance Evaluation of Color Retinal Image Quality Assessment in Asymmetric Channel VQ Coding

Agung W. Setiawan (Institut Teknologi Bandung, Bandung, Indonesia), Andriyan B. Suksmono (School of Electrical Engineering and Informatics, Institut Teknologi Bandung, Bandung, Indonesia), Tati R. Mengko (School of Electrical Engineering and Informatics, Institut Teknologi Bandung, Bandung, Indonesia) and Oerip S. Santoso (Institut Teknologi Bandung, Bandung, Indonesia)
Copyright: © 2013 |Pages: 19
DOI: 10.4018/jehmc.2013070101
OnDemand PDF Download:


The RGB color retinal image has an interesting characteristic, i.e. the G channel contains more important information than the other ones. One of the most important features in a retinal image is the retinal blood vessel structure. Many diseases can be diagnosed based on in the retinal blood vessel, such as micro aneurysms that can lead to blindness. In the G channel, the contrast between retinal blood vessel and its background is significantly high. The authors explore this retinal image characteristic to construct a more suitable image coding system. The coding processes are conduct in three schemes: weighted R channel, weighted G channel, and weighted B channel coding. Their hypothesis is that allocating more bits in the G channel will improve the coding performance. The authors seek for image quality assessment (IQA) metrics that can be used to measure the distortion in retinal image coding. Three different metrics, namely Peak Signal to Noise Ratio (PSNR), Structure Similarity (SSIM), and Visual Information Fidelity (VIF) are compared as objective assessment in image coding and to show quantitatively that G channel has more important role compared to the other ones. The authors use Vector Quantization (VQ) as image coding method due to its simplicity and low-complexity than the other methods. Experiments with actual retinal image shows that the minimum value of SSIM and VIF required in this coding scheme is 0.9940 and 0.8637.
Article Preview


One of the most important information on the retinal image is the blood vessels. Through the retinal blood vessels, the ophthalmologist can diagnose several diseases, for example: micro aneurysms (tiny dilations of the retinal capillaries), haemorrhage, soft exudate, hard exudate, and neurovascularisation. Figure 1 shows the RGB color retinal image, the image of the channel R, G, and B. If we look at the retinal image in each channel, we can get that the green channel shows the best contrast of the blood vessels.

Figure 1.

RGB color retinal image and its channels

Explanation of why the image of the channel G has the most contrast of blood vessels is due to the green light, the absorption coefficient of hemoglobin is at the peak of the spectrum. As a result, blood vessels containing hemoglobin that absorbs more green light than the surrounding tissue. This tissue was dark, so the blood vessels will be seen more contrast in image taken with green light. A little red light is absorbed by the pigment in the eye, and red light dominates the spectrum of the reflected. This is why color fundus images appear reddish. Because the absorption coefficient of the red light is less absorbs by the inner eye than green light, the R channel image is not too much contrast as the G channel image. Nevertheless, this does not mean that there is no useful information on the R and B channels (Thomas Walter et al., 2007).

Figures 2-4 show a color coded image of the retina with an asymmetric coding scheme at each channel. In Figure 3, the number of bits in the channel R = 0.5 bits/pixel (bpp), G = 0.125 bpp, and B = 0.5 bpp, and the total bits in Figure 3 is 1.125 bpp, the value of PSNR = 40.61 dB. While Figure 4, the number of bits in the channel R = 0.125 bpp, G = 0.5 bpp, and B = 0.125 bpp. So that the total bits in Figure 4 is 0.75 bpp, the value of PSNR = 39.93 dB. We can see that the PSNR value of Figure 3 greater than Figure 4. From visual observation (subjective assessment), it can be seen that Figure 3 has a distortion, a greater blocking effect than Figure 4. From another point of view, we can compare the total allocation bit in Figure 3 and 4, Figure 4 has less bits than Figure 3, but from visual assessment, we can see that coded image in Figure 4 has less distortion than Figure 3. (Setiawan et al., 2008).

Figure 2.

Original color retinal image scaled down to 12% view

Figure 4.

Image compressed to 0.75 bits/pixel, PSNR = 39.93 dB

Figure 3.

Image compressed to 1.125 bits/pixel, PSNR = 40.61 dB

From the preliminary result, we can see that the PSNR value, as objective assessment, does not correlated with retinal image distortion. So, we have to find another method to assess the retinal image distortion. Although PSNR is a common method to assess the signal (image) quality, this signal (image) quality assessment is not in accordance with the Human Visual System (HVS). The PSNR compare the difference (error) between two signals. The HVS is sensitive with structural information, while the PSNR doesn’t able to represent distortion in structural information. Therefore, Zhou Wang et al developed a method that is able to adapt the HVS called the Structural Similarity (SSIM). (Zhou Wang, et al., 2004).

Complete Article List

Search this Journal:
Open Access Articles: Forthcoming
Volume 8: 4 Issues (2017)
Volume 7: 4 Issues (2016)
Volume 6: 4 Issues (2015)
Volume 5: 4 Issues (2014)
Volume 4: 4 Issues (2013)
Volume 3: 4 Issues (2012)
Volume 2: 4 Issues (2011)
Volume 1: 4 Issues (2010)
View Complete Journal Contents Listing