Cauchy with whale optimizer based eagle strategy for multi-level color hematology image segmentation

Pathological color image segmentation is an exigent procedure due to the existence of imperceptibly correlated, and indistinct multiple regions of concern. Multi-level thresholding has been introduced as one of the most significant image segmentation procedures for pathological analysis. However, finding an optimal set of threshold values is an extremely time-consuming task, and crucially depends on the objective function criterion. In order to solve these problems, this paper presents a multi-level hematology color image thresholding approach with the assistance of a two-stage strategy called Eagle Strategy coupled with Whale Optimization Algorithm (ES-WOA), analyzing the performance over five well-known objective functions, namely; Kapur's entropy, Fuzzy entropy, Tsallis' entropy, Otsu's method, and Cross entropy. A rigorous comparative study is performed among classical WOA and existing eagle strategy based optimization algorithms, considering a set of hematology color images, and common performance indexes evaluated by each objective function tested. Experimental results indicate that proposed ES-WOA in combination with Tsallis' entropy outperforms the rest of tested algorithms in terms of computational effort, image segmentation quality, and robustness. For example, ES-WOA with Tsallis' produces segmented images with average Peak Signal-to-Noise Ratio (PSNR) values 16.0371, 17.9975, 21.1353, and 23.0759 for threshold values 2, 3, 4, and 5, respectively, which are superior to other tested methods. Additionally, the numerical results are statistically validated using a nonparametric approach to eliminate the random effect in the obtained results.

Automated summary

This study introduces a multi-level hematology color image thresholding method based on the Eagle Strategy coupled with Whale Optimization Algorithm (ES-WOA), analyzing its performance over five well-known objective functions. The proposed ES-WOA with Tsallis' entropy outperforms other algorithms in terms of computational effort, image segmentation quality, and robustness.