A region merging image segmentation algorithm based on boundary extraction. The steps of the algorithm comprise gradient map calculation, boundary extraction, initial segmentation and region merging, wherein the initial segmentation can be omitted. In the process of region merging, the ratio of lengths, belonging to a boundary extraction result part, of closely adjacent edges of regions is taken as a merging cost, and the regions are merged according to an ascending order of average gradient values within the regions, and a texture difference evaluation mechanism is also introduced to remove false segmentation. By means of the algorithm, the problems of the current commonly used segmentation algorithm having too fine a segmentation granularity, being susceptible to noise, being susceptible to light, needing to manually mark a large number of samples, requiring a large amount of computation, consuming a large amount of memory, etc., are solved. In addition, all regions or all categories achieve an optimal segmentation effect on a final segmentation result. These advantageous features can reduce the computational resource consumption of subsequent tasks and improve the processing effects thereof.
展开▼