A complex-valued encoding golden jackal optimization for multilevel thresholding image segmentation

Multilevel thresholding image segmentation is not only an indispensable component of image computation and machine vision but also a fundamental element of image analysis and feature extraction, which has received extensive attention from many domestic and international academics in recent years. However, the threshold level rises in direct proportion to the computational complexity. Therefore, this paper presents a complex-valued encoding golden jackal optimization (CGJO) established on Kapur's entropy to accomplish this issue, and the intention is to split the available image into a multitude of conspicuous salient portions that illustrate concepts and procedures of the pertinent object. The golden jackal optimization (GJO) employs the cooperated foraging of the jackals to imitate prey searching, enclosing and pouncing to generate the optimum solution. The complexvalued encoding employs the diploid's notion to alter the real and imaginary of the search agent, which amplifies the information multiplicity and reinforces the inherent parallelism to motivate productive search and inhibit voracious convergence. The functionality and viability of the CGJO are confirmed by comparison with BWOA, DOA, COA, LCA, OOA, SABO, GJO and CRWOA. The experimental results reveal that the CGJO explores and exploits to acquire a superior convergence speed, greater numerical precision and stronger segmentation quality. Additionally, the CGJO offers excellent practicability and stability to address image segmentation successfully.