4.2. Examples
Fig. 5 and Table 1 show one example, where Fig. 5A illustrates an original image. After global BCV thresholding, the binarized image is given in Fig. 5B. The largest white object is given a label number 0. Because the object No. 0 has an area greater than 6400 and the range of gray levels is 116, the algorithm does BCV thresholding in the corresponding area of the original image, and the binarized image is shown in Fig. 5C. In Fig. 5C object No. 1, 2, 6 and 9 have an area greater than 1225 pixels. Expect for the object No. 9, the other three objects had a shape factor S>2.8, and where selected for further processing and tests. Out of these three objects, No. and 2 had a range of gray levels over 50. At the end of the whole sequence, the algorithm performed BCV thresholding both for object No. 1 and 2 in the original image, and the final result is displayed in Fig. 5D. To see if the new algorithm works better than some standard local/adaptive thresholding techniques, in the next section we will compare two widely-used thresholding algorithms.