findcircles

This function provides an example of feature-extraction in a binary image. This example could be adapted for other specific requirements.

Features are assumed to be made up of ones, against a background of zeros. The program makes an effort to remove outliers and stray pixels (line 47). Features are isolated using MATLAB's morphological routine BWLABEL. Conceptually, a square is drawn around each separate feature, then the feature is compared with a circle occupying the same square. If it resembles the circle to within a certain tolerance (line 40), the feature is counted as a circle. Features are not considered if they are smaller than a preset size (line 43). The program also finds the distance between the centers of the first two circles found. Circles may be missed if they are only partly in the image (i.e. at the edges).

NOTE: Requires IMCIRCLE.M, also provided in the MATLAB image userfiles.

The files have been tested on Matlab v. 5.3.1, and 6.0.