Image Thresholding by Improved Convergent Floating Thresholding Method Using Extended Control Value

摘要

In this paper, the authors propose a threshold setting method for binary images. This method consists of two steps which are different threshold calculating algorithms. The first step is an improved model of the Convergent Floating Thresholding Method, with switching between two control values to update a tentative threshold. By this step, thresholding of a boundary area of gray level change is performed. The second step is composed of the discriminant analysis method and is applied to flat areas of gray level. A feature of the proposed method is comparison of the gray level with two final tentative thresholds which are obtained using two coefficients as control values. With this method, it is possible to decrease thresholding error which occurs in the case of low contrast and brightness change. Pattern in similar gray-level areas can also be clarified. The proposed method and the conventional methods were tested, using general outdoor scene images. The results showed that the proposed method can output good-quality binary images and is useful for thresholding of scene images.