Research On Super-resolution License Plate Reconstruction Technology In Video Sequences Based On Image Analysis And Processing System

Monitoring equipment plays an important role in crime prevention,deterrence,warning and fighting,which has become an important tool for public security work.However,limited by many adverse conditions such as the recording circumstance,the equivalent performance and the shooting distance,the clarity of acquired images are sometimes low.In order to get clearer images,improving the performance of monitoring devices is one direction,super-resolution reconstruction of inferior images is another important direction,among which the super-resolution reconstruction of sequential images is the focus and difficulty.In this paper,the domestic police software is taken as the experimental platform for the study of reconstruction of the license plates in videos.In the process of reconstructing samples,the reconstruction quality of images with loud noises is poor.Besides,since samples are selected from moving vehicles,the images need to be restored after image reconstruction.The software needs relevant parameters for deblurring operation,and the artificial judgment of parameters are inefficient.Based on the above finding of defects,the super-resolution reconstruction algorithm and the calculation of motion fuzzy parameters are studied.Aiming at the defects of Police software,the super-resolution reconstruction system and the motion fuzzy image restoration system are designed based on MATLAB software.In super-resolution reconstruction system,POCS and blind super-resolution reconstruction algorithm are applied,which have improved the reconstruction quality of images with loud noises.In the motion fuzzy image restoration system,motion fuzzy angle and distance are calculated.In the experiments,sequential images with simulated noises and real videos are reconstructed respectively,which results verify the effectiveness of the proposed methods.The experimental results are as follows:1.Taking Police software as the experimental platform,techniques and means of super-resolution reconstruction of sequential images are studied and experiments are carried out.It is found that when the noise intensity reaches 30 dB and above,the reconstruction quality of the image by Police software is significantly decreased.Noise has a serious impact on the results of the reconstruction by Police software,which makes the extraction of license information failed frequently.In the deblurring operation of moving images by Police software,the motion fuzzy parameters need to be observed and input manually,which leads to large errors for the observed parameters,thus affects the quality and efficiency of the restoration results.2.In order to solve the problem of poor reconstruction quality of images with noises,simulation experiments were carried out on MATLAB software.In the experiments,POCS algorithm and blind super-resolution algorithm were applied to the super-resolutionreconstruction system.Taking sequential images with varying intensities of noises as samples,Police software,bilinear interpolation algorithm,POCS algorithm and blind super-resolution reconstruction algorithm were used respectively for reconstruction.The results show that for those without noise,the reconstruction quality by Police software is better,while with noises increasing,blind super-resolution reconstruction algorithm can better eliminate the impact of noise.3.Aiming at the problem of motion fuzzy parameter identification,a motion fuzzy restoration system is designed on MATLAB platform.By plotting the Fourier spectrum of a motion fuzzy image,the motion spectrum angle is calculated through Radon transform.On the basis of the calculated angle,the motion fuzzy distance is calculated through black band algorithm.Error analysis of the calculated results is carried out in experiments to verify the credibility of the system.4.In the study,videos from real cases are selected as samples for the experiments.The super-resolution reconstruction system and the motion fuzzy image restoration system were applied to the real-case videos in the experiments,which results verify the validity and practicability of the two systems.