A Noise Analysis Method Based on Full Link

Abstract

The invention discloses a noise analysis method based on the full link, which includes: calling a statistical model of the scene in the frequency domain corresponding to the image product generated by the full link of remote sensing imaging; analyzing the statistical model of the scene in the frequency domain to obtain The total power of sub-sampling noise represented by the standard deviation of the scene; according to the pre-established relationship between the mean value of the scene and the standard deviation of the scene, the total power of the shot noise represented by the mean value of the scene is converted into the total power of the shot noise represented by the standard deviation of the scene; The sum of the total power of sub-sampling noise represented by the standard deviation of the scene and the total power of the shot noise represented by the standard deviation of the scene is taken as the first total power of noise; based on the total power of the first noise, the whole chain Road noise analysis, get the analysis results. The invention realizes the quantification of noise, ensures the image quality of the overall design of the camera, and determines the algorithm and parameters for image processing.

Description

technical field [0001] The invention belongs to the field of overall design and image technology of aerospace optical remote sensing cameras, and in particular relates to a noise analysis method based on a full link. Background technique [0002] In the development of optical remote sensing imaging systems, users often restrict the SNR (Signal to Noise Ratio) parameters of the imaging system. The most commonly used is to propose the SNR index in the case of high, medium and low uniform radiance, such as the maximum SNR greater than 48dB , the medium SNR is greater than 37dB, and the minimum SNR is greater than 20dB. [0003] On-orbit SNR is determined by many influencing factors in the imaging process, including optical remote sensor detector noise, circuit noise, scene shot noise, quantization noise, compression noise, and image processing noise amplification. The existence of noise causes the annihilation of useful information in the image, the target is lost, and the amo...

AI Technical Summary

Problems solved by technology

The existence of noise causes the annihilation of useful information in the image, the target is lost, and the amount of information is reduced

[0004] Because domestic satellites mostly pursue high MTF (Modulation Transfer Function, modulation transfer function) design, resulting in sub-sampling of the imaging system, sub-sampling causes the high-frequency components of the scene to be confused with low-frequency components, causing false targets to exist, resulting in image distortion

However, in the overall design of the aerospace optical remote sensing imaging camera, the sub-sampling effect is not considered as noise, only a qualitative understanding, and in the overall design, it is very important to understand the size of the amount and its proportion in the noise

Images