Image deblurring aims to recover the clear image from the damaged image. The most existing blind image de-blurring approaches only consider estimating the blur kernel in the gray domain. In fact, for the color image produced by the digital camera, the blur effects for each color channel are usually different. This paper proposes an new approach to acquire more precise blur kernels to reconstruct image by estimating blur kernels through RGB channels independently instead of just using the gray domain. The numerical results demonstrate that proposed approach can achieve better performance compared with other state of the art methods.