Research On Motion Blurred Restoration Of Road Traffic Sign Image

In the process of acquiring road traffic sign image in car video system, due to the highspeed of the vehicle, motion blurs often appear in the acquired images. In order to implementidentification and analysis of the road traffic sign image effectively, motion blur restoration ofroad traffic sign image is needed. Image motion blur restoration is an optimization processwhich makes full use of “motion blur” priori knowledge, through the certain processing,estimates the closest to the original clear image from motion blurred image. The effectivetranslation to the priori knowledge “motion blur” is to estimate the blurred angle and blurredlength from motion blurred image.This paper mainly studies the blurred angle estimation algorithm based on gradient imagespectrum, derives the corresponding relationship between the parallel dark stripes in thegradient image spectrum and blurred angle, obtain the blurred angle estimation from theinclination angle of parallel dark stripes. The experimental results show that estimation absoluteerror of image spectrum blurred angle based on the gradient is within1degree for more than85%of the test images. The image spectrum Radon transform in the estimated blurred angle is atwo-dimensional diagram, Its main peak width is the width of the parallel dark stripes of thespectrum center. The estimated value of the blurred length can be obtained by the proportionalrelationship between the blurred length and the width of the parallel dark stripe.The point spread function can be constructed with blurred angle and blurred lengthestimate. Then motion blurred image restoration using the Wiener filter、least squares filter andLucy-Richardson filter are implemented. The MATLAB simulation experiments in this papermainly are focused on three important road traffic signs: Warning sign、Prohibition sign andMandatory sign. Automatic motion blurred restoration of road traffic sign and noisy image arerealized in the GUI interface. Improve the image resolution effectively, conduce to the furtheridentification and analysis of road traffic sign image.