A hybrid flower pollination optimization based modified multi-scale retinex for blood cell microscopic image enhancement

Abstract

Multi-Scale Retinex (MSR) algorithm enhances images that are taken in nonlinear lighting conditions. In this paper, we propose an automated approach for image enhancement using MSR and Flower Pollination Algorithm (FPA) to select the optimal weights to the different scales of Gaussian filters from the desired image for MSR. The experiments are carried out using blood cell microscopic imaging to investigate the MSR and FPA. The proposed method are compared against the state-of-the-art swarms algorithms; Particle Swarm Optimization (PSO), Artificial Bee Colony (ABC), Cuckoo search (CS) and standard MSR in the aspect of the mean, standard deviation (SD), peak to signal-to-noise ratio (PSNR) and the root mean square error (RMSE). The experiment results showed that the proposed hybrid algorithm proves itself to be robust and effective through experimental results and outperforms the state-of-the-art algorithms.