I want to convert color image into binary image with smoothing

herewith I have attached output of my code

img_orig = imread(filename); %color fingerprint image

subplot(2,3,1);

imshow(img_orig);

img_grey=rgb2gray(img_orig);

subplot(2,3,2);

imshow(img_grey);

J1=edge(img_grey,'canny',0.1);

subplot(2,3,3);

imshow(J1);

I want to retrieve smooth image like this.

so that I can process it furthur.

thank you in advance

Answer 1

[Image Analyst]

What does your output image look like? Is it skeletonized? With your "smooth image" above, they're probably have to skeletonize it first before proceeding so, if the ridges look good, then you're all set. If the ridges look bad maybe you can improve it by changing the edge detection parameters. Like try Sobel, or some ridge finding filters like Frangi filter, Hessian filter, or COSFIRE-B. I think the File Exchange has some of those filters.

Be aware that in some places the contrast is so low that essentially no fingerprint ridges are to be found.

See fingerprint papers in Vision Bibliography for much better algorithms than either you or I could think up:

Papers on fingerprint analysis

[amrin shaikh]

my output image is shown in subplot-3 will be visible when you scroll right (in image attached ove here).

I need your help because i want to process and as you can see my individual ridge is represent two lines and they are very thick i want to merge it with parallel ridge and treat them both as one.

please help me it furthur

[Image Analyst]

The only other things I can suggest, since fingerprint analysis is not one of my specialities, is to try anisotropic diffusion to help smooth the original image in the direction of ridgelines. See attached demo.