An unsupervised approach to determination of main subject regions in images with low depth of field

In this paper, we propose an unsupervised approach to separate focused main subject regions from defocused background. This algorithm first computes the blurring level using the bivariate kurtosis of all 8 times 8 DCT blocks of a photographic image with low depth of field. Then these blocks are clustered to blurry regions and sharp regions. The sharp regions are considered the main subject regions. This is a fast unsupervised approach to detect the main subject regions in photographic images with low depth of field. Experimental results show that the presented method provides higher speed than the multiresolution wavelet-based segmentation method.