Energy curve and histogram thresholding implementation and evaluation with Aquila Cross-entropy optimization algorithm

Abstract

This work proposes the implementation of minimum cross-entropy thresholding based on the aquila optimizer (MCE-AO) for the segmentation of base images and support to visually differentiate the different tissues present in the region. As an alternative to segmentation, the energy curve of the images was used, the energy curve has interesting properties, since it considers the spatial contextual information of each image and not only the intensity of the pixel as the histogram does, and in contrast to the histogram, the energy curve seems to be smoother by preserving the valleys and peaks. A comparison was made between the results calculated by the Aquila optimizer algorithm histograms of each image and the energy curve calculated for each image, showing considerable improvements in mean fitness. The quality of the segmented images was evaluated with the PSNR, SSIM and FSIM metrics, with the histogram and the energy curve, and a comparison of each of the results was made, showing improvement in the quality of the images segmented with the energy curve.