Optimal Multilevel Image Threshold Selection Using a Novel Objective Function

Image thresholding is a reputed image segmentation process, extensively used to attain a binary image from a grey scale image. In this article, a bi-level and multi-level image segmentation approach is proposed for grey scale images using Bat Algorithm (BA). In this work, two novel Objective Functions (OF) are considered to obtain the optimal threshold values. The proposed segmentation process is demonstrated using six standard grey scale test images. The performance of the proposed OF-based segmentation procedure is validated using the traditional Otsu’s between-class variance. The performance assessment between the proposed and existing OF is measured using well-known parameters, such as objective value, Peak Signal to Noise Ratio (PSNR), Structural Similarity Index Matrix (SSIM) and CPU time. Results of this study show that the proposed OF provides a better objective value, PSNR and SSIM, whereas the existing OF offers faster convergence with a relatively lower CPU time.