In order to solve the problems of insufficient scene adaptability and poor real-time performance of common automatic white balance (AWB) methods, an adaptive AWB method based on color channel histogram reconstruction is proposed, and the hardware circuit of the proposed algorithm is implemented by field programmable gate array (FPGA). It not only corrects the white balance of the image, but also ensures the high-speed real-time image processing of the system. In this method, firstly, the image is processed by restricted limited adaptive histogram equalization (CLAHE) to improve the image contrast, then the channel partition statistics of gray level interval is carried out for the image, and the reconstruction method of color histogram matching or translation is adopted for the images of different scene categories. Experimental results show that compared with AWB algorithm based on light source estimation, the accuracy of color temperature correction of the proposed algorithm is improved by 14%, which has better adaptability to different color scenes and real-time processing ability.