AN EFFICIENT MODIFICATION OF GREY WOLF OPTIMIZATION USING CUCKOO SEARCH, LEVY FLY AND MANTEGNA ALGORITHM FOR REAL-TIME IMAGE PROCESSING APPLICATIONS

Abstract

Optimization methods, frequently used in several image and video processing algorithms for the attainment of optimal solutions, pose severe hurdle in case of real-time processing. For catering to the needs of real-time operations in a cost-effective way, dedicated hardware is inevitable. The huge computational load of any optimization method strikes down its feasibility of being realized in terms of dedicated hardware. The computational complexities of meta-heuristic optimization methods are even more than any other conventional optimization methods. So, in spite of having the capability of providing global solution by dodging local optima, meta-heuristic optimizations are avoided in real-time systems. To overcome the bottleneck, in this article, a modified GWOA (modified grey wolf optimization algorithm) is formulated by blending the advantages of CS (cuckoo search), Levy fly (LV), and MA (Mantegna algorithm). This modified GWOA is articulated to be computationally efficient and precise, so that, it can easily be realized in terms of dedicated VLSI architecture while maintaining the accuracy at a high level. The proposed method helps to diminish the cost and power requirement of high end and costly real-time imaging/ video processing systems while upholding its precision. The proposed method is tested by using MATLAB R2018b. The high-level synthesis (HLS) tool of Xilinx Vivado 18.2 software is used to synthesize this MGWOA, thus establishing the viability of the proposed algorithm to be implemented on FPGA/ ASIC.