Research On Noise Model And Noise Suppression Method Of CMOS Imaging Devices

Imaging system has been widely used in the astronomy, aerospace, agriculture, industrial, biomedical and other fields. As a core part of the imaging system, the performance of image sensors has become the focus of attention. Now, there are mainly two kinds of image sensors widely used in imaging application:CCD (Charge-Coupled Device) and CMOS (Complementary Metal Oxide Semiconductor). In general, compared with the CCD image sensor, CMOS image sensors have many advantages, such as low static power consumption, strong anti-interference ability, high integration and low cost. However, CCD image sensor is better than CMOS image sensor in the noise. Therefore, if we can keep the original advantages of CMOS image sensors, and reduce the noise of the CMOS, CMOS image sensors will be more advantageous. Therefore, noise suppression in the CMOS imaging device is particularly necessary.I studied the classification of noises, the causes of all kinds of noises, the nature and manifestations of the noises in the CMOS imaging device, all of which provide a solid theoretical foundation for subsequent work. In order to propose the effective method of noise suppression, for LUPA4000CMOS imaging systems example, the study used modules which are in Simulink blocks library to create the imaging device model successfully in Simulink. On the basis of the understanding of noises, I created thermal noise, reset noise, fixed pattern noise models on the device model, and then made a simulation and detailed analysis of the imaging device model and noise models. In addition, the research achieve correlated double sampling technique in the two kinds of low-speed and high-speed in the device model. In the high-speed system, the effect of noise is relatively large, so I proposed the thinking of correlated four sampling technique, and made a simulation, verification in the imaging device model, and gave the comparison and analysis. The idea of creating models has laid a foundation for follow-up proposing the method, and also provided a new research idea for other engineering practice.Finally, given the noise theory and the creation of the noise model and imaging device model, I learned that the fixed pattern noise is the largest noise in the imaging device, so the paper proposed a method to suppress fixed pattern noise, and used the M language to implement this method in the MATLAB environment, which were applied to the image taken LUPA4000CMOS imaging system, and achieved a good result. The paper gives comparison charts of the method, one of which is the images before the noise suppression, and other is after. This paper also analyzed the inhibitory effects, calculated the size of the fixed pattern noise and improved the percentage of before and after noise suppression.