Comparison of Otsu's and Rosin's Methods for Threshold Determination

Abstract

Real time edge detection of objects in images implemented in FPGA circuits is often based on gradient calculation using appropriate convolutional masks. The resulting edge map is obtained by thresholding the gradient image. The key issue is determining the threshold value. In practice, Otsu's and Rosin's methods are most often used to determine the threshold value. Using images from three databases, we compared these two methods and the results are presented in our paper.