数字视频编解码技术国家工程实验室深圳分室
National Engineering Laboratory for Video Technology – Shenzhen Division

图像去模糊

For motion blur

 

1. We propose a method to remove motion blur from a single photograph by using priors.

2. We use a no-reference image quality assessment method [1] to evaluate our deblurring method and compare it with several state-of-the-art methods.

3. We collect many real blurry images which are shown in testset, our results are shown in result. All the blurry images are taken by cellphones or digital cameras.

 

For defocus blur

 

1. We propose a method to estimate the defocus blur map, and compare the results with several blur map estimation methods.

2. We propose a single image deblurring algorithm to remove spatially variant defocus blur based on the estimated blur map.

3. Some real blurry images taken by cellphones are shown in test_set.

 

Download: paper   MATLAB code   results   dataset

 

 

Reference

 

[1] Y.Liu, J.Wang, and S.Cho, “A no-reference metric for evaluating the quality of motion deblurring,” ACM Trans. Graph, vol. 32(175), pp. 1–12, 2013.