Application Of Non-reference Image Quality Evaluation In License Plate Screening

License plate screening can be seen as a pretreatment process for license plate recognition.By filtering out the license plate images that can be identified and analyzing the images that cannot be identified,the license plate image acquisition system errors are corrected to achieve the purpose of improving the accuracy of license plate recognition.In this paper,a no-reference image quality assessment algorithm based on image information entropy is used for license plate screening.The characteristics of the license plate image acquisition system are briefly analyzed,and it is concluded that the license plate image quality evaluation algorithm needs to cover the distortion types such as noise,blur,and block effects.According to the influence of image quality on the accuracy of license plate recognition,the screen should reflect the brightness,sharpness and contrast of the image.The comparison of the brightness,sharpness and other aspects of the four existing license plate image quality evaluation algorithms is compared.It is found that none of the three algorithms gives the overall quality of the license plate image,and the other is a simple weighted sum of the four evaluation factors.According to the shortcomings of these algorithms,a license plate image quality evaluation algorithm based on image information entropy is proposed.The spatial entropy,spectrum entropy,and gradient entropy of the license plate image are used as feature vectors of the image,and the overall quality score of the image is calculated using the two-stage framework structure.The spatial entropy value reflects the brightness of the image,the spectrum entropy value reflects the sharpness of the image,and the gradient entropy value reflects the contrast of the image.An interactive and friendly man-machine interface was designed and used to detect the performance of the algorithm.This algorithm can guide the MCS hill-climbing method and one-way search algorithm to find the right aperture and focal length by evaluating the brightness and sharpness of the image.That is,by analyzing the image with poor quality,the error of the image acquisition system can be corrected.This algorithm can filter out images whose image quality is less than the threshold value in the image database,and the accuracy,consistency and monotonic performance are better than those of the algorithm compared to the article.It means that the quality of the image can be accurately screened out.