Integration of local information-based transition region extraction and thresholding

Transition region-based thresholding method utilizes the pixel location and neighborhood information to segment an image into a number of interesting regions which have specific characteristics in recent years. In this paper, a novel transition region extraction and thresholding method is proposed, which is based on the integration of the weighted local entropy with the improved local grayscale difference. The integration of two modified local information can character the intrinsic quality of transition regions easily and effectively. For some synthetic and real images, the proposed method is quantitatively and qualitatively compared with other transition region-based thresholding methods such as local entropy method, gray level difference method, modified local entropy method, and as well gray-level histogram-based thresholding methods e.g. Otsu-based method and entropy-based method. The experimental results have confirmed the validity and efficiency of the proposed approach.