Research On Computer Image Mosaic Technology For Low Illumination Environment

Image stitching refers to the process of registration and fusion of a group of overlapping pictures in the same scene to generate a new image containing all the information of the original image,taking into account the high-resolution and wide-field parallel requirements.At present,most mature image stitching technologies are based on clear,easy-to-process images collected under good lighting conditions,while image stitching technologies in low-light conditions such as morning and dusk,and night are not yet perfect.Night-time low-illumination image stitching technology has more urgent practical significance in the field of security monitoring.Since night is a high-frequency period during which a security event occurs,it is extremely important to achieve scene image stitching in low-illumination environments such as night.The images collected in low-illumination environments are relatively noisy.These noises interfere with the true pixel values of the image and are easily amplified when the image is enhanced.Denoising processing effectively removes image noise while preserving image details.In low-light environments,the contrast of the image is low,and the texture information is small,which is not conducive to the extraction of features in subsequent image registration,which affects the quality of image stitching.This paper uses the dark primary color prior dehazing technology to enhance low-illumination images,and according to the difference between foggy images and low-illumination images,adaptively corrects the transmittance according to the estimated atmospheric light constant to achieve better enhancement effects.In order to apply the ORB algorithm to night image stitching,it will be disturbed by lighting factors,and it has the characteristics of high local contrast and relatively concentrated feature points.A histogram equalization algorithm is used to enhance the low-contrast areas of the image and equalize the histogram The two sets of features extracted before and after are used to calculate the perspective transformation matrix,thereby obtaining a more accurate perspective transformation matrix,optimizing the stitching effect,and finally using the asymptotic and fading position weighted average fusion algorithm to complete the image stitching.