Abstract: Relevant research on image restoration indicates that image's subjective visual quality is closely related to its local details. A novel restoration algorithm for defocus blurred image was proposed. The proposed algorithm based on BTV regularization framework by introducing a local adaptive weighted function constructs a new cost function for imargc restoration. hhis cost function which not only takes into account the global data-fidelity, but also considers the local statistical properties of image,meaning to fully consider the local structural features of image under global data-fidelity,hence behaves much better in edge preservation. Experimental results confirm the effectiveness of the proposed method.The image is restored with better subjective and objective visual quality, compared with other methods such as traditional B丁V regularization approach.

Key words: Image restoration, Point spread function, Bilateral total variation, Local weighted function