Study And Implementation Of Automatic White Balance And Exposure Algorithms In HD Camera

Recently, digital camera is becoming more and more popular, due to its simple operation and the convenience of getting pictures once photographed. If digital image preprocessing ensure the high quality of the image, then the three kinds of preprocessing automation system in digital camera –AWB, AE, AF contribute much to the shooting facilitation of digital camera, and also have extremely significant effect on the final image quality. This paper mainly researches on the automatic white balance and automatic exposure in digital image preprocessing.In terms of automatic white balance, we elaborate the basic principle and carry out careful research on several common white balance algorithms such as gray world, perfect refection algorithm and integrated arithmetic. and points out the deficiency of these algorithms after evaluation and analysis. Combined with laboratory hardware conditions for the project design, an implementation mechanism is put forward, which can obtain better processing effect in the condition of smaller resources utilization. The implementation is based on the hardware and software, and the hardware part is responsible for raw data statistics in real-time image with Verilog language, software part responsible for calculating the equation in DSP with C language. Then we simulate various scenarios with our scheme to evaluate automatic white balance effect.In terms of automatic exposure, the algorithms for automatic exposure was analyzed in detail, including the average brightness algorithm, the weight-mean algorithm, the algorithm based on brightness histogram and the algorithm based on image entropy. According to the shortcomings of the existing automatic exposure algorithm based on image entropy in the judgment of optimum exposure time, the increment setting of exposure time and the searching for peak area, this paper presents an improved automatic exposure algorithm based on image entropy. We also have several experiments to evaluate the effect of exposure under different scenarios, proving that the improved algorithm performs better and is suitable for different scenarios. Besides, the improved algorithm can exposure more quickly and accurately compared with the original algorithm based on image entropy. Ultimately the improved algorithm based on image entropy is proved to be an effective and reasonable AE algorithm.