Restoration of Partial Blurred Image Based on Blur Detection and Classification

Figure 5

The estimation result of the cepstrum based estimation method. (a) is the original image used to test the cepstrum based blur kernel estimation algorithm. (b) is the defocus blur scale estimation result. The test images are generated using the original clear image and the defocus blur kernels whose scale ranges from 6 to 20. The red line is the defocus blur scale of the test images; the green line is the estimated blur scale using the cepstrum based algorithm. (c) is the result of the motion blur scale estimation results. The test images are generated using the original clear image and the motion blur kernels whose scale ranges from 6 to 20. The angle of the motion blur kernels is set at 30 degrees in order to test the motion blur scale estimation algorithm. The red line is the motion blur scale of the test images; the green line is the estimated motion blur scale. (d) is the motion blur angle estimation results. The test images are generated using the original clear image and motion blur kernels whose angles range from 10 to 80 degrees. The blur scale of the motion blur kernel is set at 8 pixels in order to test the motion blur angle estimation algorithm. The red line is the motion blur angle of the test images, the green line is the motion blur angle estimation algorithm based on cepstrum, and the blue line is the improved estimation algorithm based on Radon transformation.

(a)

(b)

(c)

(d)