Automatic White Balance Algotirhm Based On Network Camera

With the rapid development of industrial automation and network, the demand for network cameras is growing. Usually when a camera take pictures, the color of the captured image may change with the changing of the light source. In high color temperature light source, the captured image will be bluish while in low color temperature light source it will be reddish. It is an urgent task that how to restore the color of the image captured by camera accurately and efficiently under different light source. In this thesis, we propose a method for white point detection based on the auto white balance technology. Meanwhile, based on the proposed method a self-adaptive auto white balance algorithm which can be used in camera is realized.Firstly, we introduce the relevant knowledge of image and video signal processing and provide the procedure of image signal processing in ISP. Then we describe the related theory about image white balance, including auto white balance, color temperature character and the concept of RGB/YUV color space character. On this basis, a more detailed exposition of some of the current auto white balance algorithm is provided. In the current algorithms, there exists much shortage. So, an improved auto white balance algorithm is proposed based on the white point detection method and the procedure that how to implement the auto white balance algorithm in camera system is described. Finally, detailed analysis of the implement procedure of software algorithm and the hardware is described. The detailed test results of the proposed auto white balance algorithm are as follows: in the light source of 2400K/3200K/3550 K color temperature, the central value of saturation is below 80mv/50mv/50 mv after the auto white balancing. And the values of the color saturation [red(103±13 dec), yellow(55±21 dec),cyan(95±40 dec) ] and phase [red(122±13°), yellow(168±11°),cyan(294 ±23°) ] are all in the design range.