Noise Analysis And Image Processing Of CMOS Image Sensor

With the development of large scale integrated circuit manufacturing and design technology, CMOS image sensors with low cost, low power consumption, high integration, etc., and close to or above the level imaging CCD image sensor, has a very broad application prospects.At first,in this thesis we analyzed the analog noise of CMOS image sensor, mainly based on time and space noise, this thesis described three-dimensional noise model. On the ground of it we designed a three-dimensional noise measurement test system. And then we capture a series of pictures of homogeneous background, after analysis and process the pictures, we can get the value of time and space noise under different illumination, and reached its trend.The digital noise of CMOS image sensors will be generated in digital-analog conversion. In order to minimize the noise of digital image, it's necessary to deal with the image collected by CMOS image sensor at the bank-end. Except median filter and Gaussian filter, identification part of the noise was added to program. According to the method,the image was processed in simulated environment MATLAB. First the type of noise was indentificated, then process it According to the results of last step. And this method shows a very good effect on noise minimization by the experiments.In practical application, especially under the situation of darkish light, the imaging system will take the method of adding additional light source to increase exposure. In this paper, the inhomogeneity of the optical system, especially the inhomogeneity that the light caused, asymptotic stun in the optical system, dark current in the imaging system and heterogeneity of the light response were discussed. This thesis uses the calibrated matrix method for calibration, and this method shows a very good correction effect by the experiments.