Parameter estimation method of electron multiplying CCD (Charge Coupled Device) noise model

Abstract

The invention discloses a parameter estimation method of an electron multiplying CCD (Charge Coupled Device) noise model. The method comprises the steps of: first, initially setting a noise distribution model as a mixed Gaussian distribution model to process; then, carrying out maximum likelihood iteration on the noise distribution model, substituting the set initial value to the mixed model to solve a posterior probability density of a sample value from a Gaussian source, and then substituting potential data to a log function with incomplete data to calculate a partial derivative to solve an extreme value so as to obtain an iteration estimated value of the parameter; and finally, judging and comparing the iteration estimated value with the initial value; judging whether the terminating condition is met or not according to a circulation terminating condition, if so, stopping iteration; if not, setting the iteration value as the initial value, and carrying out maximum likelihood iteration again. According to the invention, maximum likelihood estimation of the noise parameter of the electron multiplying CCD image is realized by simple steps, so that the complexity of the maximum likelihood method is effectively reduced, and the electron multiplying CCD image noise can be quickly and accurately estimated.

Description

technical field [0001] The invention belongs to the electron multiplication CCD image processing technology, in particular to a parameter estimation method of the electron multiplication CCD noise model. Background technique [0002] Electron multiplying CCD adds a first-stage gain register on the basis of the traditional CCD structure to achieve on-chip gain, suppresses the readout noise that dominates in traditional CCDs, and has the advantages of low noise, high sensitivity, and high dynamic range. It has broad application prospects in fields such as night vision, astronomical observation and biomedicine. Many domestic research institutes and universities have invested in the research work of electron multiplication CCD. Wang Mingfu, Xie Zongbao and Bu Hongbo from the Institute of Optoelectronic Technology of the Chinese Academy of Sciences, the Shanghai Institute of Technology of the Chinese Academy of Sciences, and the Beijing Institute of Space Mechatronics conducted ...

AI Technical Summary

Problems solved by technology

According to the noise source and statistical characteristics of the electronic multiplication CCD, the noise distribution model of the output image of the electronic multiplication CCD is a mixed Poisson-Gaussian noise distribution model, and the parameters of the mixed Poisson-Gaussian distribution model are estimated using the traditional maximum likelihood estimation method , there is a problem that the maximum likelihood function is difficult to solve

Images