Automatic White Balance Algorithm Research And Hardware-Software Implementation

As electronic techniques is continuous improved rapidly cameras or video camcorders used for image retrieval technology and development become digitalized. Color cast phenomenon is a major difficulty encountered by digital image processing. According to the principle of black body radiation, the same pixel has different colour under different color temperature environment. The color of photographs would look very different due to differences in light projection illumination when we take pictures using digital camera. Human eyes are able to automatically adjust the color when the illuminations of the light source vary. Howerer, the most frequently used image sensor, CCD or CMOS sensor, can not correct the color as human eyes. Therefore, the automatic white balance algorithms have emerged which is a key step of image processing in digital camera. If a digital camera does not have white balancing function, a captured white object will appear reddish when under a low color temperature light source and bluish when under a high color temperature light source.First of all, this article introduces knowledge of digital image processor(ISP) and image-video signal processing, gives image processing flow chart in ISP. Then leads to the main content of this paper.Secondly, this article describes theoretical basis of image white balance: color constancy, color temperature and von kries model. On this basis, this article introduces some of the existing automatic white balance algorithms and analyzes the advantages and disadvantages of each method.Thirdly, for the shortcomings of existing algorithms, taking complexity of algorithm and effect of white balance into account, this article raises two kinds of automatic white balance algorithms. Experimental results show that both algorithms have achieved better results, and both do not need to store the image data to complete white balance calculations.Fourthly, this article discusses the implementation of automatic white balance algorithm on the system. Detailed introduces the division of hardware and software and its implementation process. Verilog hardware description language is used in hardware. Software is realized in the 32-bit MCU CK610. Both completed the process of automatic white balance. Fifthly, this article describes the system verification of automatic white balance and describes vefication process of hardware part. Briefly introduces the debugging and ISP system and gives the analysis results and automatic white balance. Finally, the articlae proposes direction for further study and how to improve the performance of the algorithm.