need a help about a paper

hi. im working on a paper(i attached it)in part III.D of it it works on a mask. it makes a mask and adds it somehow to binarized image. i wan to know what really it doest and how to makes that. i just need a litle guidance and explanation. i want to know how they really used the mask on the image. if you dont have time i explain it here:

The proposed mask consists of a circular mask inside and a ring mask outside shown in Fig. 4(a). The region 1 is a circular mask (denote as reg1) with radius r1=wmi d-1. The region 2 (denote as reg2) is a ring mask with the inner radius r2=wmid+1 and outer radius r3=2wmid. We calculate the average gray value S1 in reg1 and S2 in reg2 of the two masks separately when the center of the mask moves along the key points. The points between the two masks are eliminated for edge points are located in this region and edge points’ smoothness may cause errors in comparison. First, we determine if the key point cen- tered in (x0, y0) belongs to the inner point or the outer point. Under the assumption that the detecting points are brighter than the background, S1>S2+t1 indicates it is the inner point and then we turn to process the pixels in the mask with Alg1. Oppositely, S1<S2+t2 is treated as the outer point and we turn to Alg2. Alg1: this process is for inner points. Fig. 4(b) displays the mask moving to the inner points. Generally, two parallel edges are retained in the edge map. However, if one side is missing in the edge detection step, we can still fi nd the inner points according to the distance rela- tionship. For pixels in the ring mask, their values are lower than the pixels in the circular mask. It means that if the average gray value of reg1 is less than the chosen threshold T1, the whole region including the circle surface is classifi ed as part of the characters. Alg2: it is the operator for outer points. When the center of the mask moves to the outline of the characters (Fig. 4(c)), parts of reg2 are lo- cated in the detecting character. And reg1 of the whole circular mask is counted as background pixels. Compare the gray value of each pixel in the ring mask with S1

[Geoff Hayes]

moeen - once you have chosen the file (to attach) you must press the Attach File button so that it is attached to your question.

[moeen kamali]

thank you @geof hayes so much

[Jan]

It is unlikely that the voluntary participators of the forum read your papers and spend the time to explain it. I suggest to ask a specific question, because replying to "need a little guidance" is very demanding.

Answer 1

[Image Analyst]

They're using a circular mask. Please see the FAQ: http://matlab.wikia.com/wiki/FAQ#How_do_I_create_a_circle.3F

Also see my attached circle masking demo.

[moeen kamali]

thank you so much image analyst for your answer but i want to know how really they used the mask on image

[Image Analyst]

Sorry, I can't take the time to delve into the details of the algorithm in the paper. http://matlab.wikia.com/wiki/FAQ#Can_you_program_up_the_algorithm_in_this_article_for_me_and_explain_it_to_me.3F

Answer 2

[yusuf]

need a help about a paper circle detection

[Image Analyst]

If you have an image of the paper circle, you can post the image in a new question. If the circle is on a background of some contrasting color then you might be able to find the circular/disc-shaped paper in the scene by thresholding, just like I do to find the circular coins in my Image Segmentation Tutorial.