In this paper, an image-segmentation algorithm using a new gray level thresholding method is proposed in badly illuminated images. Badly illuminated images in general do not have an optimal unique thresholding value. For these images, it is desirable ...
In this paper, an image-segmentation algorithm using a new gray level thresholding method is proposed in badly illuminated images. Badly illuminated images in general do not have an optimal unique thresholding value. For these images, it is desirable to determine multiple-thresholding values in semented sub-regions. However, in sub-regions which are included inside objects or background, thresholding values can not be directly determined because of an absence of thresholding information. The proposed algorithm first obtains thresholding values in sub-regions where edge information can be provided. Then thresholding values in sub-regions where edge information is absent are determined by using an iterative processing. In this way, a thresholding map, which has information of thresholding values in each sub-region for an original image, is obtained. In the proposed iterative process, statistical information such as the mean and standard deviation of gray level in sub-regions are used. Finally the segmented objects are obtained from an original image using the thresholding map. Experimental results indicate the effectiveness of the proposed algorithm.