Research On Image Acquisition And Recognition Processing Technology Under Strong Light Background

With the rapid development of digital image and machine vision technology,real-time image acquisition and enhancement equipment has been widely used in intelligent transportation,intelligent agriculture,community security and other industries.The images collected under complex strong light background usually have serious problems of strong light transition area,local highlighting halo and high contrast,which seriously affect the visual sensory experience of the subsequent image content and the efficiency and accuracy of computer recognition.Therefore,this paper mainly focuses on the rapid method of image noise suppression and content enhancement under complex background.First of all,the current mainstream image denoising and enhancement algorithm theory is analyzed and studied,its main architecture is deduced,the connection and difference between different algorithms are analyzed,and the core ideas of each algorithm are extracted,which provides a theoretical basis for the realization of the method proposed in this paper.Secondly,by studying the image characteristics under strong light background and the actual effect of existing image enhancement algorithms,combined with the deficiency of Retinex algorithm in light transition region and strong light region prone to halo phenomenon,a Retinex image enhancement algorithm based on adaptive segmentation is proposed.The improved watershed segmentation based on the depth of field was used to segment the image into light and dark regions.The light and dark regions were enhanced with Retinex in different scales and directions respectively.Then,according to the edge characteristics of the light and dark regions,the fusion weights were calculated and the fusion reconstruction of the light and dark regions was carried out to obtain the enhanced image.The proposed algorithm has a good enhancement effect in the experiments of high light and shadow transition and low illumination backlight image enhancement.It can suppress the halo in the transition region and enhance the details in the highlight region,and uses less computing resources than the traditional multi-scale enhancement schemes.Then,in view of the widespread phenomenon of local strong light and halo in nighttime images collected by vehicle-mounted devices,a brightness attenuation model of local strong light images based on guided filtering enhancement is established based on the proposed Retinex image enhancement algorithm based on adaptive segmentation.The model separately attenuates the intensity of the intense halo region to weaken the influence on the display effect of other regions in the image,and then performs weighted fusion of the processed results with pixel distance as the dominant factor,and then strengthens the edge details of segmentation through guided filtering.Experiments show that this model can effectively reduce the brightness of local strong light area in low illumination image and improve the overall visual comfort of the image.Finally,an image processing simulation platform was built to compare and verify the segmentation based image enhancement algorithm,and to quantify the time efficiency and enhancement effect of the proposed algorithm and the current mainstream algorithm.The local brightness attenuation model based on guided filtering enhancement is compared in the actual environment.Experimental results show that the proposed algorithm achieves a good balance between time efficiency and image enhancement.There are 43 figures 5 tables and 71 references in this paper.