Research On Gray Level Error Conpensation Method For Optical4f System

Optical information processing is implemented based on optical spectrum analysis,fourier synthesis technology, and spatial filtering technology. The advantages ofoptical information processing system are high speed, parallelism and interconnection.Optical4f system is one of the important implementation systems of opticalinformation processing. Due to the optical device and experimental environment, thereexists noise interference in optical4f system. The noise mainly include random noise,coherent noise and gray level error, which will decrease precision of opticalinformation processing based on optical4f system. This paper mainly studies the graylevel error conpensation method for optical4f system.Gray level error compensation method based on histogram matching is researchedin this paper. The gray level error compensation image is obtained utilizing histogrammatching according to the histogram of input image. The experimental results showthat the proposed method achieves1.75dB averagely PSNR gain for the gray levelerror compensation image, the histogram matching method can compensate the graylevel error for optical4f system effectively.Gray level error compensation method based on histogram matching and radialbasis function neural network is researched in this paper. The nonlinear transformationof histogram between input and output images in optical4f system is fitted by radialbasis function(RBF) neural network, then the optimal estimation of curve forhistogram matching between input and output images is obtained. The gray level errorcompensation image is obtained utilizing histogram matching according to the optimalestimation of curve for histogram matching. The experimental results in optical4fsystem show that the proposed method can compensate the gray level error in optical4f system effectively and the visual effect of images processed is improved. Theproposed method achieves2.038dB averagely PSNR gain for the gray level errorcompensation images. The optical experimental results verify the proposed method.There is no need to know the histogram of input image in advance for this method, sothis method can be used widely.Gray level error compensation method based on histogram matching and supportvector machine is researched in this paper. The nonlinear transformation of histogrambetween input and output images in optical4f system is fitted by support vector machine(SVM), then the optimal estimation of curve for histogram matching betweeninput and output images is obtained The gray level error compensation image isobtained utilizing histogram matching according to the optimal estimation of curve forhistogram matching. The experimental results show that this method can compensatethe gray level error in optical4f system effectively, and the quality of output image isimproved. The proposed method achieves2.169dB averagely PSNR gain for the graylevel error compensation images.