Automatic White Balance Algorithm Research And Hardware Implemention

The human visual system has the characteristics of color constancy,so people don't think the color of the object changes when the irradiation light source on the object changes within a certain scope.But image sensor doesn't have the characteristics of color constancy in the process that the optical image convert to electrical signals,which would make the picture taken under the influence of light source color.In order to eliminate the influence of light source color,appropriate white balance technology is adopted to correct the color cast.The common automatic white balance algorithm based on histogram is failure when there are many high saturation pixels.An automatic white balance based on histogram matching is applied.Increasing effective judgement about the white balance algorithm based on histogram matching according to the distribution characteristics of histograms,we adopt histogram translation algorithm when the histograms have a large number of high saturation pixels.The combination of histogram matching and histogram translation algorithm makes up for the deficiency of the two algorithms effectively.In order to measure the adjustment effect and hardware resource consumption of algorithm proposed better,we completed the hardware implementation of the algorithm in this paper.In the design process,we implement the histogram matching agorithm with the of accumulationmatching parallel method in order to reduce the area of memory and operations.By using SMIC 65 nm CMOS technology,The working frequency can achieve 150 MHz,the synthesized design area is 95850um2,the power consumption is 6.37 mW.This work has obvious advantages inThe correction effect,besides compared to the traditional automatic white balance algorithms,this work reduces the input to output delay,only occupies one clock cycle;Compared to other automatic white balance based on histogram,this work takes advantage on both the correction effect and range of application,especially the range of application increases by 13.8% with the hardware resource only increased by 2.4%.