The Research On HDR Image Synthesis And Display Algorithm For Moving Object Photographing

Moderate lighting is important for photos captured by the camera.But the dynamic range of natural scenes far exceeds the hardware limitations of standard digital camera sensors.In this case,the generated photo will contain over-exposed areas and under-exposed areas,which cannot sense the scene details.One solution is to combine multiple low exposure dynamic range(LDR)images with different exposures into a single high dynamic range(HDR)image.However,the motion of objects in the scene is unavoidable,which will result in ghost-like artifacts in the fusion result.When a complex non-rigid moving object appears in the scene,it is very difficult to solve the artifact problem using conventional alignment algorithms.In order to solve the above problems,the author has done a lot of literature reading and research on the existing HDR synthesis and ghost removal algorithms,understand the principles of various methods,weigh the advantages and disadvantages,and finally choose an algorithm framework that directly eliminates ghost effects.Based on the framework,this paper proposes its own innovative improvement methods for motion pixel detection,image fusion,merging of dynamic and static regions,and processing of saturated regions of dynamic pixel regions.The details are as follows:1.Aiming at the problem that the existing motion pixel detection method takes too long,this paper proposes to add a SLIC superpixel segment pre-processing step be-fore implementing motion pixel detection.A dynamic area detection method based on IMF was subsequently designed.This improvement converts pixel-level detec-tion to region-level detection,reduces time loss,and at the same time makes the detection results of dynamic pixel areas appear as concentrated large areas,which is conducive to subsequent Poisson editing of dynamic and static areas.2.A screening method for multi-exposure image sequence reference pictures is de-signed.This method focuses on the standard deviation parameters of the whole image.This screening strategy efficiently selects the best reference picture.3.Research the image exposure fusion algorithm and understand the principle.Based on this,two modifications to the exposure fusion algorithm are proposed.Later experiments and objective data analysis prove that one of the modified methods of the weight function can obtain a HDR image with greater contrast,and the effect is better than the classic fusion algorithm.4.Analyze design flaws and make up for them.In particular,to compensate for the loss of texture information caused by overexposure in the dynamic region of the image,this paper proposes a texture enhancement method based on the Retinex principle.This method uses Alpha blending to fuse the illuminance component of the reference image with the preliminary HDR image results.Get a better HDR image.5.Design comparative experiments to obj ectively analyze the differences between this design and three other classic HDR synthesis algorithms in recent years.And sum-marize the advantages and disadvantages of this design over the other three designs.