Research On Image Super-resolution And Noise Reduction Technology Based On Area Analysis

With the continuous development of communication and information technology,the application of digital images and video is becoming more and more extensive.However,due to the defects of imaging equipment,bad weather,insufficient lighting and other factors,the resulting image often has problems such as insufficient resolution and noise,which will affect the visual effect of the image and the extraction of effective information.Therefore,research on super-resolution of noisy images will become more and more necessary.In this paper,based on the example learning model of the SI(Super Interpolation)algorithm,a super-resolution algorithm based on texture features is first proposed.This algorithm uses the texture features of the image to cluster the image blocks,and is completed by offline training and online reconstruction similar to the SI algorithm.In the whole super resolution process,this algorithm has a better super resolution effect than the SI algorithm.Then,a noise reduction algorithm process based on difference curvature is proposed.The difference curvature feature can distinguish the smooth point area,noise area and edge point area of noisy images.This feature is used to cluster image blocks,and then through the same offline training and online reconstruction to achieve a noise reduction algorithm based on example learning.This algorithm is slightly less effective when used alone,but it is used in combination with a super-resolution algorithm based on texture features,can get a better super-resolution effect of noisy images.Based on this,a new type of noise reduction super-resolution interpolation structure is proposed,which combines the two algorithms into one and realizes super-resolution of noisy images.Modelsim simulation was carried out on the proposed super-resolution algorithm with noisy images,and the implementation process of each module of the algorithm was analyzed and explained in detail.Then,by comparing the hardware simulation results of the noisy image super-resolution algorithm with the software implementation results,the accuracy of the hardware simulation is verified.A small high-definition camera system is designed.The overall introduction of the camera system from the aspect of data processing.Because the noise reduction process of the proposed noisy image super-resolution algorithm must be applied to the three channels of YUV,it will consume a lot of hardware resources and is not friendly to small hardware systems.A bilateral filtering algorithm based on Bayer template is proposed and implemented.In order to obtain a better imaging effect,an improved histogram equalization expansion algorithm is proposed.Finally,based on the system,the video resolution of 640 × 360 @ 50 Hz is improved,and a super-resolution video image of 1280x720 @ 50 Hz is generated.