Improved Radon Transform and Differential Autocorrelation Based Motion Blur Parameter Estimation Method

Restoring of motion blurred image is an important research direction in digital image processing technology, which is of great importance in many fields including aerospace, remote sensing technology, medical image processing and so on. However, the existing motion fuzzy parameter is still inaccurate when solving the length and angle imprecisely, which impact the restoration of motion blurred images. The paper proposes improvement methods for this problem. First, Radon transform is accepted in measuring relative motion angle in image pre-processing. Least squares line fitting based accurate angle estimation algorithm is proposed for small blur length. Second, an improved method based on differential autocorrelation length estimation is obtained for high accurate relative motion length estimation. In addition, this paper proposes the corresponding motion fuzzy parameter estimation algorithm for the images with different fuzzy degrees to make the estimation results more accurate.Experimental results demonstrate that the proposed method can estimate motion blur parameters more accurately. The restored image is superior to the existing methods in both subjective and objective. The proposed method is of great importance in engineering.