Thresholding Techniques Comparsion on Grayscale Image Segmentation

Image segmentation is a prime domain of computer vision backed by a huge amount of research involving both image processing-based algorithm and learning-based techniques. Due to this there is an upsurge in different segmentation technique from the research community. Various image segmentation techniques have their strength and weakness and some specific application are more geared up to some segmentation techniques. The automation systems like object detection, robotics and intelligent video analytics, do a lot of segmentation technique and hence there is need to evaluate the performance of these techniques. The paper implements the different types of segmentation techniques. Threshold techniques including like histogram thresholding, mean thresholding, edge thresholding, variable thresholding and percentile (P%-tile) exists. Algorithms are applied using MATLAB coding on the considered images. White Pixel Ratio (WPR) parameter is used for analysis of various methods. Similar to the theoretical concept, practical approach shows that WPR is better for histogram thresholding as compared to other techniques.