Research On Backlight Compensation Coprocessor Based On Hardware Acceleration

Backlight compensation which is one of the key processing technique to improve the digital image quality makes images taken in the backlight comfort human eyes, smoothness and naturalness at the same time by mainly modifying the brightness of the backlight image.Taking images is generally under front light, but it is in the backlight environment in some cases, then the objects photographed just block the light source, which leads the objects to be dark due to underexposure, but the brightness of the background around the objects is higher than the objects’, thus the backlight image has formed. Because human eye only can distinguish the outline of the object in the backlight image, but they can’t recognize the detail and texture of the object, so it is necessary to compensate for images. With the development of science and technology, the much higher request is bring up for the quality and real-time of images. As the real-time images processing system based on software has low processing speed, so the image processing based on hardware acceleration has become a research hotspot. Now Field Programmable Gate Array(FPGA) which has fast speed and good portability is the programmable logic device used widely, and it is suitable to simple algorithm and large-scale image processing. This paper designs a hardware architecture of backlight compensation algorithms based on FPGA and DSP. The main research works in this paper are as follows:1. Based on the further research on the existing backlight compensation algorithms, the backlight compensation algorithms are divided into four types:(1) the backlight compensation method of using post-processing technology(2) the backlight compensation method of using clustering(3) the backlight compensation method of combining with AE(4) the backlight compensation method of processing the chromaticity characteristic value of the image. According to different categories, there are four backlight compensation algorithms to be introduced in detail in the paper, and their performance, advantages and disadvantages are also analyzed.2. After synthesizing and measuring the performance and complexity of many backlight compensation algorithms, this paper uses the algorithm named that adaptive height-modified histogram equalization and chroma correction in YCb Cr color space for fast backlight image compensation. The algorithm based on global histogram equalization is the backlight compensation method of using post-processing technology, whose advantages are fast processing speed, adaptability and good performance, so it is suitable to the hardware design with real-time processing. The paper uses this algorithms to design the structure based on FPGA and DSP in detail, and the hardware design has the following modules:(1)converting RGB to YUV(2) the histogram statistics(3) the AHMHE algorithm(completed in DSP)(4) the color correction(5) converting YUV to RGB, and the design ideas and steps of the five modules are also given in detail.3. Finally this paper simulates and analyzes the BLC design simulation and provides the corresponding simulation diagram. Simulation analysis shows that within the valid range, the design can compensate the brightness of images taken in the backlight environment, improve backlight image subjective quality and make the processed image meet human comfort. At the same time, this paper also makes a resource evaluation to the hardware design of the algorithm, and the evaluation results indicate that the design in the paper has these advantages including saving hardware resources and good real-time.