White Blood Cell Detection and Counting based on Genetic Algorithm

This paper proposes an approach for detecting and counting white blood cells (WBCs) by taking the power of Genetic Algorithm (GA) and Otsu thresholding method. This technique is based on image binarization in HSV color space using Otsu thresholding method and applying GA for output improvement. In fact, Otsu uses HSV values in order to detect WBCs which causes false detection because of HSV values' similarity. There are some intensity values in RGB color space which corresponding HSV values are similar. This similarity makes mistake in WBC detection. Thus, GA removes false areas based on within-class variance reduction as a fitness function. This combination causes output improvement in order to maintain correct areas as WBC. Experimental results show the superiority of this approach over the comparing methods.