Research And Implementation Of Automatic Exposure And Automatic White Balance Algorithm Based On UAV Camera

In recent years,the UAV industry is booming.UAVs with high-definition cameras using for Aerial photography,tracking and three-dimensional reconstruction are spring up.In addition to a stable and reliable flight control system,a real-time and efficient camera system is also essential for UAV to complete various tasks.This paper mainly discusses and studies the camera system from two aspects: the choice of hardware and the design of image.For hardware selection,depending on the work environment and the way the photo is taken of UAV,we choose the image processing and image acquisition module mainly from the aspects of weight,volume,consumption,price and so on.And then we complete the the assemble and optical correction of the camera.For image preprocessing algorithm,this paper mainly study and implement automatic exposure and Automatic white balance algorithm.In terms of automatic exposure,we analyse several existing algorithms in detail.Such as Luminance Mean algorithm,Weighted Mean algorithm,Histogramming algorithm and an algorithm based on Information Entropy.These algorithms some are applicable to a small situation,some are difficult and complex.So this paper proposes first decimate the image and then expose it,further more,adds a termination factor for the algorithm based on information entropy,whice reudces the iteration number of the algorithm.Finally,we compare this new algorithm with these old algorithm,and find that the new algorithm is feasible.In the aspects of auto white balance,we divide the algorithm into two parts—the detection and correction of Color deviation.After Color deviation Detection,these normal Image no longer need to carry out Color deviation Correction.It is necessary to put the color deviation detection before correcting the color deviation.For color deviation decetion,we propose an algorithm,which bases on YIQ color space.The color deviation factor of the whole image is calculated by calculating the mean value and variance of the chroma of I and Q channels.With this factor we can divide those images to be tested into normal pictures and color offset pictures.However,the algorithm can not accurately distinguish the true color offset image from the essential color deviation image.To improve the accuracy of the algorithm,we propose a method,which based on histogram,for the second detection.Color offset correction algorithm is an improvement to the Gray-World algorithm.First of all we split the picture into a block of equal size,and then we get a weight coefficient for each small piece.This algorithm also eliminate these large single color block,which can cause a serious impact on the effect of color offset correction.Then,we verify the algorithm from the aspects of subjective and objective.And finally,it is proved that the improved algorithm can greatly improve the accuracy of color offset detection and the effect of color offset correction.In the last place,this paper analyzes the cemera from software system.Through the PC serial port tool we verified that the communication between camera modules is normal.And compared with the results of Matlab simulation,the image preprocessing algorithm is suitable for this camera.