A Watershed Segmentation Algorithm Based on Ridge Detection and Rapid Region Merging

Due to the effects of noise, quantization error and regional texture details, traditional watershed algorithm tended to lead to over-segmentation. Especially, when the ridges of image are of discontinuity and the edges are of fuzziness, the over-segmentation is significantly worse and the blob of image is easy to lose vital boundary. In this paper, in order to solve these drawbacks, a watershed segmentation algorithm based on ridge detection and rapid region merging was proposed. This algorithm reconstructed discontinuity ridge using ridge historical information which has reserved the information of image segmentation after opening and closing operation and completed pseudo-blobs marks based on Bayes rule, and achieved the mergence and elimination pseudo-blobs by the way of the rapid merging of the maximum similar region. Test results showed that the new image segmentation algorithm could solve the problems of over-segmentation and under-segmentation to the greatest extent, which could greatly increase the accuracy of segmentation of metallographic microgram.