Abstarct: Defocuing can be caused by different reasons mainly limitation of focus depth of camera lens. A map which indicates defocus values for image pixels is referred to as “defocus map”. The diversity in imaging condition and subject in the scene cause numerous challenges in the problem of estimating the blur map. In this thesis we specifically investigate the problem of color similarity between background and foreground. In this thesis we propose a method for estimating defocus map of color images. In this method first defocus map is estimated for image edges. Then image is segmented to superpixels baxsed on color consistency. For each superpixel the defocus value is estimated as the median of defocus for edges inside the superpixel. To improve the consistency of defocus map within a local neighborhood we use color attribute and a detail measure, baxsed on singular value decomposition, for each superpixel. The result of evalution for the proposed method shows that the proposed method provides better estimation for defocus map in compare with other methods, espcessially for images in which focus and defocus areas have similar color. The proposed algorithm was also applied on an image dataset to segment images as focus and defocus regions, the MSE and PSNR criterions for this experiment are 0.2219 and 7.5858 respectivly. The proposed algorithm is considerably faster than other methods.