Research On Key Technologies Of Image Enhancement For Post Inspection

In the customs security inspection,animals and plants are listed as prohibited items due to the possibility of carrying viruses.At present,the main detection of this kind of prohibited items is X-ray security inspection machine,which can detect the items without opening the luggage package.It is widely used in various fields of transportation and is crucial to our travel safety.However,the security inspection images mainly adopt the method of manual recognition,and the detection efficiency is very low.Therefore,it is necessary to enhance a large number of security inspection images that are difficult to recognize,so that the security inspection personnel can quickly and accurately identify the items.Security inspection images generated by X-ray transmission often have the disadvantages of large noise,unclear edges,and low contrast.Targeted enhancement algorithms include histogram equalization,wavelet transform,and contourlet.This article studies the dual-energy X-ray security inspection system images.Different colors can be added according to different attributes of objects to form pseudo-color images,which is more convenient for security personnel to identify.Since the above enhancement algorithms are based on processing grayscale images,they are not suitable for the images studied in this paper.Based on the research of dual-energy X-ray security inspection system images,this paper proposes the following two algorithms to solve the above problems:adaptive histogram equalization in the spatial domain and improved Retinex image enhancement algorithm in the frequency domain.Analyzed their respective applicable images,and further proposed separate enhancement algorithms based on different color differences.The specific research contents are as follows:(1)Aiming at the security image with partially blurred details,in the spatial domain enhancement algorithm,an adaptive histogram equalization enhancement algorithm based on guided filtering and LoG operator is proposed.First use guided filtering for denoising,then use the LoG operator to detect the edge of the image and enhance it.Finally,the brightness component in the HSV color space is limited contrast adaptive histogram equalization processing to enhance the image contrast and reduce color distortion.(2)For security inspection images with low global contrast,the Retinex algorithm is improved in the frequency domain enhancement.First,through experiments,it is found that the multi-scale Retinex algorithm with color recovery has obvious enhancement effect on the security inspection image,but there will be a color shift phenomenon,so an improved Retinex algorithm is proposed.Then,referring to the method of expanding high-frequency components in the Unsharp Masking algorithm to enhance the edges,the high-frequency components are enhanced in the improved algorithm,and the image contrast is adjusted with Gamma correction.(3)Through analysis,the spatial domain method proposed in this paper is suitable for images with low local contrast,and the frequency domain method is suitable for images with low global contrast.Security inspection images include both low local contrast and low global contrast.Therefore,it is proposed to classify the images according to the size of the average value of the pixel difference,and respectively adopt suitable algorithms to improve the detection efficiency.The experimental results and data show that the algorithm of this paper has achieved the expected effect on the security image enhancement,and the research on the security image enhancement in this paper has certain use value.