Research On The Technology Of High Resolution Infrared Image Acquisition And Recognition

In order to detect CMYK metameric image,a miniaturized and portable infrared image acquisition and recognition system is designed.The hardware part of the system integrates the main control module of the Raspberry Pi,the image acquisition module of the infrared CCD camera and the image display module.The software part realizes the discrimination and enhancement algorithm of infrared image definition,and integrates the ZBar two-dimensional code recognition algorithm.In order to solve the problem of low quality of infrared image,based on the former,this paper uses Android mobile phone instead of LCD screen,socket communication + Base64 coding technology to realize wireless transmission of infrared image,and zxing algorithm to recognize two-dimensional code.For low-quality infrared images(low resolution,low contrast,uneven illumination),the related software algorithms are studied to solve the problem of low-quality infrared images.The research work of related software algorithms is as follows:(1)In order to solve the problem that the clarity of the infrared image collected by the camera fluctuates greatly in the process of focusing,several clarity discriminant functions,such as gray gradient,are implemented in this paper.The analysis and comparison show that Laplace gradient function has the characteristics of stable monotonicity and more sensitive to the clarity change of the image,which is conducive to accurately screening the infrared image with relatively high clarity.(2)Aiming at the problems of low contrast and uneven illumination in the collected infrared image,this paper adopts a homomorphic filtering algorithm based on the combination of CLAHE and Laplace sharpening.The infrared image is processed by homomorphic filtering to make its brightness distribution uniform.The contrast of the infrared image is improved by CLAHE processing,and the edge of the infrared image is enhanced by Laplace sharpening.At the same time,the algorithm can be applied to real-time processing of infrared image acquisition and recognition system.(3)Aiming at the problem that the resolution of infrared image is generally too low,this paper designs a depth residual network algorithm based on three-layer jumper.The network consists of input convolution module,output convolution module,multiple residual modules and two sub-pixel convolution modules.In this paper,a residual error module with double-layer jumpers is designed to strengthen the correlation of feature information between image convolutional layers,and the Adam algorithm is used to train and optimize the network to obtain a reconstruction model of infrared images.When the low resolution image is input into the network model,the extracted residual image will be superimposed with the low resolution image,and the high resolution infrared image will be obtained after being amplified by the sub-pixel convolution module.By analyzing the experimental results,compared with bicubic,srgan and srresnet,the infrared image reconstructed by this algorithm has higher PSNR and SSIM,and better visual effect.