Abstract: Addressing the issues commonly encountered in traditional threshold segmentation methods, such as susceptibility to local optima, inefficiency, and high sensitivity to noise, this paper introduces a modified strategy integrated grey wolf optimizer (MSI-GWO), which is applied to threshold image segmentation based on the minimum symmetric cross-entropy. The algorithm incorporates an improved Tent chaos initialization to enhance global search capability and accelerate the optimization process. It modifies control parameters to help the population escape local extrema. Furthermore, the integration of a random walk strategy significantly improves the efficiency of searching for the optimum solution. Verified by six standard test functions, MSI-GWO shows superior convergence performance compared to traditional intelligent optimization algorithms. When applied to threshold image segmentation based on minimum symmetric cross-entropy, MSI-GWO demonstrates a marked improvement in performance metrics, such as feature similarity (FSIM), structural similarity (SSIM), and peak signal-to-noise ratio (PSNR), with an increase in the number of thresholds, confirming the application potential of the algorithm in the field of image segmentation.
Keywords: grey wolf optimizer (GWO); Tent chaos initialization; random walk strategy; minimum symmetric cross-entropy; multi-threshold segmentation
合肥工业大学学报自科版