Research And Realize Of Auto-expose\auto-whitebalance In Video Processing Front End

The video quality is mainly depended on the video processing front-end. TheAutomatic Exposure and Automatic White Balance (Auto-Exposure\Auto-WB) algorithmis the most important factor in the video processing front-end. Therefore a goodAuto-Exposure\Auto-WB algorithm is needed by video capture devices to capture a highquality video.An Auto-Exposure\Auto-WB algorithm is studied, and it has beenimplemented and verified in the actual application environment in this paper.To solve the exposing problem caused by the particular lighting conditions such asbacklight and facelight, an Auto-Exposure algorithm based on BP neural network todistinguish the lighting conditions is proposed. The lighting condition is distinguishedthrough BP neural network, and the ideal brightness value of the video image is calculatedaccording to the lighting condition. Then the brightness value of the video image is set tothe ideal value by the video capture controller. The Auto-Exposure algorithm is applied onthe platform of network camera with TMS320DM368. The experiment indicates that theexposing problem caused by the particular lighting conditions has been improved, such asbacklight and facelight.To extend the effective scope of the current Auto-WB algorithm, two new Auto-WBalgorithms are used in combination. The white pixel on the perfect reflection Auto-WBalgorithm is expanded to the gray pixel in the first Auto-WB algorithm, and thecorresponding detection standard of gray-point is presented in this Auto-WB algorithm.Amethod that uses Image entropy to delet the large color block is proposed in the secondAuto-WB algorithm. The second Auto-WB algorithm is used in case of the failure of thefirst algorithm. The combination of the two algorithms has been implemented on theplatform of network camera with TMS320DM368. The experiment indicates that effectivescope of the current Auto-WB algorithm is indeed broaden, and the performance of thealgorithm can be satisfied for the requirements of the network camera white balancefunctions.