Scene-based self-adaptive heterogeneous-fixed-noise removal method

Abstract

The invention describes a scene-based self-adaptive heterogeneous-fixed-noise removal method which is characterized by being realized through the following steps: 1. determining scene initialization; 2.scene self-adaptive iteration; 3. scene template update; 4. image generation and output. The method is applicable to heterogeneous-fixed-noise removal. The method is capable of adapting to change of the scene without generating ghost images; and the convergence speed is high so that a convergence problem and a ghost image problem of the prior algorithms are solved.

Description

Technical field [0001] The present invention relates to the technical field of non-uniform noise removal, in particular to a scene-based adaptive non-uniform fixed noise removal method, which is an adaptive non-uniformity compensation method for removing fixed pattern noise on a focal plane array . Background technique [0002] In many applications, focal plane arrays are used to capture images. For example, in a target tracking system, an uncooled infrared camera is used to capture infrared images for tracking. The focal plane array includes infrared detection units arranged in a matrix. For example, some uncooled infrared cameras have 640*480 detection units, and each detection unit outputs one pixel. Due to manufacturing process and other reasons, the response rate of each detector unit is not exactly the same, and there are certain differences between them. For example, for the same incident radiation given by each detection unit, the output pixels of some detection units ...

AI Technical Summary

Problems solved by technology

Calibration-based non-uniform background noise removal algorithms, although these algorithms are simple, they cannot solve the problem of fixed pattern noise drift, and their adaptability is very limited

Although the scene-based adaptive non-uniform background noise removal algorithm can overcome the problem of background drift to a certain extent, it always has the problem of algorithm convergence or scene ghosting, and most adaptive non-uniform background noise removal The algorithm does not give an effective initialization method. Due to the existence of these problems, the application of existing algorithms is limited.

Images