March 9, 2020 - Trying out New Color Segmentation Method Part 2

Lab Work

Last week, I began to try and find a different way to segment the histology images by color in order to find a faster and possibly more accurate way to threshold the images instead of the method that I was previously using. The new way that I found which might be useful is seen in the article “Color spaces in OpenCV (C++ / Python)”, and I began to follow the steps used in the article to segment images by color on this image:

20180924-angasi022-10x-tiled.PNG

But after looking at the article a little more, I realized that they were using mutiple images in order to compare what the color spaces look like for the same object under different lightings. Thus, I used two more images (seen below) to use as comparisons in this process:

20180924-angasi121-40x.jpg

20180924-angasi162-40x-scale.png

I compared the 3 images through 4 color spaces: RGB, Lab, YCrCb, and HSV. The results are shown below:

original: original_images.png R: r_comparison.png G: g_comparison.png B: b_comparison.png L: L_comparison.png a: A_comparison.png b: B(lab)comparison.png Y: Y_comparison.png Cr: Cr_comparison.png Cb: Cb_comparison.png H: h_comparison.png S: s_comparison.png V: v_comparison.png


Next Steps

Compare the color spaces of these images (finding similarities and differences) and continue working through the article tutorial

Written on March 9, 2020