Research On Super-resolution Reconstruction Algorithm Of Vehicle Plate Images

Vehicle plate image plays an important role in the intelligent traffic system and public security area.However,in the actual surveillance video,the low resolution of the vehicle plate image seriously affects its application because of the influence of the long distance between the car and the camera as well as the illumination.It has been one of the most urgent tasks in the intelligent traffic system that how to improve the resolution of the vehicle plate image.Image super-resolution reconstruction is a technology that improves the resolution of image by signal processing technology,and it enhances the resolution of images by algorithms.The thesis is to propose an algorithm of vehicle plate image super-resolution reconstruction based on analyzing the characteristics of the sequence of vehicle plate images.First,the initial estimation of high-resolution vehicle plate image is obtained by exploring structure similarity between input images containing the same vehicle plate.Then,the thesis studies the image sparse representation and proposes the algorithm of vehicle plate image super-resolution reconstruction based on sparse representation.The initial plate image is enhanced by exploring the common prior knowledge of plates based on image sparse representation method,and a sharp high resolution plate image is finally obtained.The proposed algorithm combines the structure similarity of sequence images and sparse representation into the process of super-resolution reconstruction,and achieves the goals of taking advantages of both algorithms and making up for their shortcomings.Finally,experiments on simulation images and actual images shows that the proposed vehicle plate image super-resolution reconstruction algorithm improves the resolution of the vehicle plate image,and achieves satisfied result visually and quantitively.