Wide Dynamic Range Algorithm Design Based On The Image Signal Processor

The wide dynamic range camera has the ability of greater dynamic range imagingthan ordinary ones. With the increasing requirement of high quality imaging, widedynamic and low light imaging have been the most important features of a camera.There are two methods to achieve the wide dynamic range feature of cameras: one isusing of wide dynamic range sensor, another one is using multi-exposure HDRItechnology. The former is strongly real-time with high cost and general effect. The laterone is weaker real-time but with low cost and excellent results. If the computationalcomplexity can be reduced when keeping the quality almost unchanged, it will help toimprove the market competitiveness of the product.In this thesis, a wide dynamic range algorithm is designed for the image signalprocessor of camera. The imaging principle of camera is served as a research startingpoint of the algorithm which is based on the multi-exposure HDRI technology and thecharacteristics of imaging signal processor. The advantage of the algorithm is to make itpossible to obtain almost the same effect when the computational complexity issignificantly reduced. First, the multi-exposure HDRI technology is introduced, andthere is a detailed analysis of the method by Debevec and Grossberg in which acquiringthe high dynamic range image by recovering the camera response function. Then a fastmethod of acquiring HDR image is proposed with an intensity relationship and animage fusing strategy based on partition. The performance of the proposed algorithm istested by imaging quality, running time, and the ability of ghost artifacts removal.