Infrared image transition region extraction and segmentation based on local definition cluster complexity

According to the problem of extracting transition region inaccurately based on typical local complexity method, due to its excessively low complexity measurement and deficient detail representation, we propose an improved infrared image transition region extraction algorithm. By constructing local definition cluster function and calculating its complexity, we improve the complexity measurement of the image to a great extent, which is able to represent more detail information. Experiments validate this algorithm. The results show that our method based on local definition cluster complexity can extract the transition region more accurate, and segments the image much better compared to typical local complexity method.