Adaptive multi-threshold image segmentation using neighborhood minimum gray values for enhanced 2D histogram construction
Yunfei Zhang, Zhan Zhang, Ri Xu, Ping Xiong · 2025
This study presents an adaptive segmentation technique employing two-dimensional entropy analysis for constructing enhanced 2D histograms. While existing approaches primarily analyze central pixel intensity and neighborhood mean values, we observe that conventional horizontal/vertical grayscale variations inadequately capture complete local texture characteristics. Our innovative solution addresses this limitation by developing a minimum-intensity based algorithm that generates 2D histograms through combined analysis of center pixel values and the minmum intensity within 3×3 neighborhood. Comprehensive validation testing performed across the complete MATLAB image repository confirms the superior performance of our proposed methodology compared to traditional approaches