Multiple contour segmentation with automatic thresholding

This research proposes a scheme to quickly search for multiple contours. By selecting only two points on the image, the initial line formed by these two points would reveal the points intersected with contours. With these points on the contour outlining the boundaries of an object, its closed-form contour would be detected. Additionally, whereas the thresholds must be set by users in other conducted researches, in the proposed work, the thresholds are automatically generated by the gray-level difference between the points on the initial line. Rather than an image having multiple independent contours or multiple overlapping contours, only two initial points are required to find the desired contours. Compared to the other semi automatic segmentation methods, human errors and operating time are greatly reduced by using the proposed scheme.