An expanded histogram approach for multilevel image thresholding

In this paper a new image thresholding technique is proposed based on expanding the histogram of the image to accommodate spatial-related information in the form of a variance map of every gray level present in the image. The expanded histogram along with the variance levels are fed into a thresholding finding algorithm based on partitioning the interval (histogram) in an optimal way using dynamic programming with an entropy-based cost function. Compared with many existing methods, simulations on a range of images show good results. The effectiveness of the algorithm is shown even in the presence of low to moderate additive Gaussian noise levels.