Research On Multi-Exposure HDR Image Generation Method

With the improvement of public living standards, visual aesthetic requirements are also improving, people has ultimate pursuit and expectations on the image quality and visual experience. High dynamic range (HDR) image has been known and used by consumers, and technologies for acquiring HDR images have attracted more and more attentions from the academic community and the industry. At present, due to the limitations of hardware, most of the captured images are low dynamic range, which are difficult to cover the entire dynamic range of natural scene. Although the HDR imaging equipment that directly capture HDR image has emerged, such equipment is expensive and difficult to popularizing in the mass market. At present, the acquisition of high dynamic range image is mainly based on the software method. The most widely used software method is to obtain HDR image by multi-exposure, which is the mainstream HDR imaging technology. Multi-exposure HDR imaging technology includes two categories: one is the method based on the inverse camera response function recovery;the other is direct fusion method. The former one often causes color shift when used in RGB color space; the fusion method only can slightly expand the dynamic range, it's not enough for generating authentic HDR image.To solve the problems described above of multi-exposure HDR imaging technique,two algorithms are proposed in this thesis. One is based on multi-resolution decomposition. The algorithm applies the inverse camera response function to restore the image radiation value, and set the weight of each pixel according to the pixel characteristic, combined with the pyramid model to implement the image fusion. This algorithm has good performance for expanding image dynamic range, preserving the details of the image, avoiding the color shift according to the subjective evaluation and the objective evaluation. Another algorithm is to generate HDR image in YUV color space. To settle the issue of color shift for generating HDR image in RGB color space,this scheme decomposes the image into luminance and color components, and uses the inverse camera response function to mapping the luminance component. Furthermore,the human visual response model is introduced to enhance the perceived intensity of the luminance channel and extend the dynamic range of the image. Experimental results show that this algorithm can avoid color shift with low computation complexity.