Advances in Natural and Applied Sciences

In this paper, bi-level and multi-level thresholding is presented for the standard RGB images using Otsu and a novel Brownian Bat Algorithm (BBA). Maximization of between-class variance is chosen as the objective function during the optimization search. The performance of the proposed BBA is confirmed by considering six benchmark RGB images and compared with the existing bat algorithms such as Traditional Bat Algorithm (TBA) and the Levy flight Bat Algorithm (LBA). The evaluation of performance between the proposed and existing bat algorithms are done using existing constraints such as objective function, Root Mean Squared Error (RMSE), Peak to Signal Ratio (PSNR), Structural Dissimilarity (DSSIM) index, and algorithm convergence. The result evident that proposed BBA offers better values for objective function, RMSE, PSNR and DSSIM, whereas TBA and LBA offers faster convergence compared to BBA.