How can I find the intensity value which has the maximum number of pixels in the histogram? I'm using grayscale image. please I need a code ..

how can find a function to solve this?

Answer 1

[Walter Roberson]

[counts, edges] = histcounts(YourIntensityImage(:), 0:255);
[maxcount, maxidx] = max(counts);
most_common_intensity = edges(maxidx);

[Ahmad Yahya]

Thank you very much .. Solved

[Ahmad Yahya]

please, could u help me with another part, with all of thanks? I'm working on a stage of image processing research and i need a matlab code to remove all the recto and verso dark text from a manuscript page with just save the watermark in this page which has the light color, may the morphological operations (closing, erosion, ..) help ?

[Image Analyst]

Do you already have the watermark removal/separation code? If not, you'd need to know the method that was used to make the watermark if you want the best possible separation. I attach a method for watermarking and recovery using LSB method. By the way, I have no idea what recto and verso mean.

[Walter Roberson]

I suspect recto and verso refers to bright and dark text, and it sounds like the watermark is a visible watermark rather than a hidden one. If so then it would be "foreground removal" question, where the foreground includes both dark and bright and the background is intermediate.

[Image Analyst]

Some watermarks, like "video bugs" are actually transparent except for a edge line (you can see the video through them but are "outlined" by an "edge"). Others are "solid" (completely cover the background video). Of course these would require different recovery algorithms.

[Ahmad Yahya]

Yes, that's what I meant sir Walter Roberson. recto and verso refers to bright and dark text, because i did an optical scanned to this manuscript and the foreground and background had overlapped. The watermark here is visible watermark. so i need to do "foreground removal" for the dark text(recto and verso texts) with keep visible light watermark showed in the sent image? Please if u can help me to find code to do that.. even if we use morphological?

[Ahmad Yahya]

Thank you for your cooperation Image Analyst.

[Walter Roberson]

We suggest posting an image; it might give us some ideas.

[Ahmad Yahya]

[Walter Roberson]

Use a mask to select the points with brightness more than a certain value. That will eliminate the text and black lines. Then you can do detection on what remains.

[Ahmad Yahya]

Could you explain more with an example please, i couldn't understand the idea . thanks

[Image Analyst]

brightStuff = grayImage > someThresholdValue;

I doubt this will produce much you can work with. It will probably be a mess.

[Ahmad Yahya]

I've experimented this and it didn't give the required result..I need to remove all the darkest text (old Arabic hand writing)??

[Image Analyst]

You might try a modified median filter, like my attached code.

Answer 2

[Image Analyst]

For a more general and somewhat different solution, you might look at the color frequency image: http://www.mathworks.com/matlabcentral/fileexchange/28164-color-frequency-image

[Ahmad Yahya]

Thank you very much for your help .