A Fast Thresholding Technique in Image Binarization for Embedded System

Abstract

In the embedded systems of visual navigation, due to the limitations of images capture working environment conditions, there are sometimes defects of nonuniform illumination and noisy in the captured images. Therefore, global threshold methods are unfeasible. On the contrary, image binarization using local threshold methods are more appropriate. However, local threshold methods which take more time to calculatet can’t satisfy the requirement of real-time performance in embedded systems. In this paper, taking these two limitations into consideration, we proposed an efficient and fast method to determine the threshold value by using integral image and statistical methods which is similar to the local threshold method and meet the requirement of real-time performance in embedded systems. After experiments, for the proposed method, the result has demonstrated that the processing speed is nearly twice the global threshold method. And the processing quality closes to that of local threshold method.

DOI : http://dx.doi.org/10.11591/telkomnika.v12i1.3359