Circular spot detection and segmentation with rotating line local binary analysis

The circular spots appear as round local peaks/valleys in image space, and they reflect omnidirectional symmetry and Gaussian distribution characteristics. According to these observations, a novel rotating line scan operator is proposed to achieve feature analysis of the spot object. Experimental tests show that the proposed algorithm is robust to spots´ appearance variation (size, local contrast and intensity). The method presents promising performance in various real images.