Research On Technologies Of Enhancement And Reproduction Of Overexposure Image Sequence

For a long time,researchers' innovative technical solutions and algorithms in the field of image enhancement are basically for dark-light images,and rarely for overexposed images.This has caused serious scarcity of this type of image enhancement technology,which means that the field of image enhancement There are still many challenges and difficulties to be solved.This paper conducts related technical research on overexposure image enhancement,proposes a more effective enhancement technical solution,and solves the problem of insufficient detail collection of low dynamic range(LDR)images through high dynamic range(HDR)images.The research results include the following work:1)Propose a single overexposure image enhancement technology based on adaptive gamma.According to the characteristics of overexposed images,overexposed images are divided into two categories: overexposed images with light source and overexposed images without light source.The quality of an overexposed image with a light source is mainly affected by its light source and halo.Therefore,an adaptive gamma enhancement algorithm based on distance and brightness is proposed for such effects.By determining the light source area,the distance from each point in the image to the light source area is further calculated and normalized.The normalized distance is combined with the brightness to design an adaptive gamma function,and the image fusion algorithm is combined to enhance image details and reduce Effect of halo,and finally restore the true color of the image through color correction.Experimental results show that the algorithm greatly reduces the effect of halo,the image details are clearer after enhancement,and the visual effect is better.For the overexposed image without light source,an improved algorithm based on the existing enhancement algorithm-an adaptive gamma enhancement algorithm based on moving window particle swarm optimization(PSO)is proposed.By moving the window,the time required for parameter learning is reduced,and the detailed information of the image in the window area is enhanced at the same time.In addition,a new fitness function is proposed,which allocates adaptive parameters for image detail information,image exposure and image contrast.Compared with the existing algorithms,the algorithm proposed in this paper not only considers the enhancement of the overall details but also the enhancement of the local details.At the same time,the moving window also greatly reduces the total time of the PSO algorithm and improves the efficiency of the algorithm.2)Propose an overexposure image sequence enhancement technique based on brightness balance.Different from single overexposure image enhancement,overexposure image sequence enhancement not only considers the enhancement effect of each frame,but also considers whether the image connection between frames in the sequence is smooth.There are complex differences in brightness between frame images,and there is also content correlation between sequence images.Therefore,if you simply use the single overexposed image enhancement algorithm,the brightness difference will still exist or even become larger,and the details of the image will be supplemented.Insufficient,making the visual effect of the enhanced sequence poor.Therefore,on the basis of the single overexposure image enhancement algorithm,an overexposure image sequence enhancement algorithm based on brightness balance under the same scene is proposed.The brightness balance algorithm is used to control the brightness continuity between the frame images,and the detail fusion between the frames is used to enhance and enrich the image details to make the overall visual effect of the sequence better.3)Proposed a filtering-based HDR image display technology.At present,the tone mapping method based on the filtering decomposition model is the most common,but these methods have insufficient mapping of details and insufficient adjustment of image brightness.Aiming at these problems,a novel and effective algorithm is proposed-an HDR image display algorithm based on retinal model and edge preservation filtering.Construct two brightness channel(cone channel and rod channel)images,use perceptual quantization curve to normalize the channel image,and use edge preservation filter to decompose the converted channel image to obtain the detail layer and the basic layer.The inverse trigonometric function is used for the detail layer to enhance the details,and the histogram adjustment based on the look-up table(LUT)is used for the basic layer to perform dynamic range compression.Finally,the two brightness channel images after the mapping are weighted and merged to obtain the final brightness image.Experimental results show that the image details are more perfect after tone mapping,the image brightness is more peaceful,and the visual effect is better.