The Hardware Implemetation And Software Verification Of Video Signal Pre-processing IP

With the popularity of portable electronic devices, the digital photography has penetrated into people’s daily life, and consumers for high quality image also promote the rapid progress of manufacturing technology of image sensor. In recent years, image sensor resolution of consumer electronic equipment is constantly being refreshed, the image sensor with 5 million and 8 million pixels has become standard in the low-end consumer electronics devices, the image sensors of high-end consumer electronics devices have up to 16 million pixels. To increase the resolution of image sensors will increase the amount of image data, in order to reduce the influence brought by increase the amount of data, a majority of image sensor manufacturers choice Bayer Raw format as the output of the image sensor, and for the cost considerations, the chip designers will tend to video signal preprocessing unit integrated in SoC.Now,the procedure of video signal pre-processing are already very mature, mainly including the gamma correction, noise removal, artifact removal, CFA interpolation, white balance, and so on. Based on the product specifications, only the gamma correction, automatic white balance and CFA interpolation will be studied and implemented in this paper. For the gamma correction, this paper introduces the concept and functions of gamma correction, studies the problems about the implementation on hardware of the index operation in the gamma correction, and then proposes a method based on lookup table and piecewise combination, which can greatly reduce the cost of hardware. For white balance, this paper introduces the concept and functions of the white balance, analyzes the gray world algorithm, perfect reflection algorithm and orthogonal combination algorithm, then an improved algorithm base on the Nakano algorithm was proposed, and uses the subjective evaluation method to evaluate the above algorithms, finally introduces the implementation method of improved algorithm on hardware. For the CFA interpolation, this paper introduces the concept and functions of CFA interpolation, and studies the classical bilinear interpolation algorithm and some algorithm joined the judgment of image edge, for example, the first-order differential edge interpolation algorithm, the second-order differential edge interpolation algorithm and the Adams-Hamilton adaptive interpolation algorithm, then makes some improvements to Adams-Hamilton adaptive interpolation algorithm, evaluates the above algorithm use the method of subjective evaluation and objective evaluation with PSNR, finally introduces the implementation method of improved algorithm in hardware.In software verification, this paper uses C and Verilog to implement algorithm at the same time, to verify by contrast to each other. In the aspect of hardware implementation, the paper implements the video signal pre-processing IP on the latest Xilinx Zynq-7020 FPGA platform, and uses AR0542 of Aptina to provide video signal input for IP.According to the final realization of the software simulation and hardware results, determine the correctness of our design, and image effect and the hardware cost can meet the design requirements.