Improved Otsu segmentation based on sobel operator

The classical maximum variance algorithm which could be called Otsu is stable, however, the threshold calculation is rather low which leads to an undesirable performance in terms of segmenting the target from the background during the strip detection process. The poor segmentation process will have negative impact for the subsequent steps, such as the target extraction and the target recognition. In order to overcome this disadvantage, we proposed an improved thresholding segmentation approach based on the maximum variance algorithm. The proposed approach consists of two major steps. The first is to adapt the Sobel operator to extract the grayscale information at the edge. The second is to apply Otsu on the extracted grayscale information. According to our experimental results, the proposed approach achieved the desirable segmentation performance.