Feature point extraction for palmline

Hie i m confused in an algorithm of palm line matching.below i m writing the step that i have to do.please, try to briefly explain this step and suggest any way to implement this .

input :- I have the location of feature point and the input image of palm line

step- for each feature point,the orientation of its corresponding palm line is computed as the direction of the line segment of length 8 which has the maximum average contrast(absolute difference value)to its immediate surrounding(parallel segments placed 2 pixels above and below the given segment)

Thanks in advance.

[Image Analyst]

What if the line is vertical? How can you inspect the pixels 2 above and below that line?

Answer 1

[Image Analyst]

Try using improfile(). Call it three times for center, and 2 pixels away and parallel to the center line. Then rotate it through several angles to find the best.

[Promila]

hey ..improfile only work for grey level and RGB images but i have a binary image.so, what can i do for this?

[Walter Roberson]

double() applied to the binary image will give you a gray level image. If, that is, your binary image is of data type logical: if it is already in a numeric data type then it is already a grey-level image.