Image strip noise removing method and system based on spectral filtering

Abstract

The invention relates to an image strip noise removing method and system based on spectral filtering. The method comprises the following steps: reading a panchromatic or multispectral image; recognizing periodic noise and bad line noise in the image; compensating the periodic noise by adopting a gain adjustment method; and compensating the bad line noise by adopting a left-right column average method. According to the invention, indexes such as detection unit response capability and detection unit response difference of an image are defined and calculated; the method is advantaged in that strip noise type determination and strip noise position quick determination can be realized, the whole processing process is automatically realized through utilizing a computer, manual intervention is notneeded, processing speed is fast, the automation degree is high, and automatic batch processing of massive images can be realized. The method is suitable for satellite images with irregular stripe noise, and can be used for quickly recognizing and removing stripe noise positions and types of different types of remote sensing images such as panchromatic remote sensing images, multispectral remotesensing images and hyperspectral remote sensing images.

Description

technical field [0001] The invention relates to the technical field of remote sensing image processing, in particular to a method and system for removing image band noise based on spectral filtering. Background technique [0002] Remote sensing images are often interfered and affected by various noise sources in the process of generation and transmission. Among them, banding noise is the most important kind of noise. Banding noise is a special kind of noise that has a certain periodicity and directionality and is distributed in bands in the image. [0003] The generation of band noise is mainly affected by the imaging mechanism of the sensor, the instrument's own factors and external factors. The essential reason is that the response capabilities of each detection unit of the detector to the received signal are not completely the same, and vertical stripes of light and dark appear on the image, which is also the root cause of the band noise. Specifically, the influencing ...

AI Technical Summary

Problems solved by technology

[0004] (1) Inhomogeneity of the detector device itself

Due to the limitations of current material manufacturing and technology, various defects such as uneven doping, unequal thickness and uneven size of detection elements will appear in the materials of CCD devices, resulting in different parameters between different detection units, resulting in differences in the response of detection units

In addition, the wear and tear of the instrument, etc. will also cause the inconsistency of the detector response

[0005] (2) Influence of CCD dark current

[0006] (3) Inconsistency introduced by working status

During the working process, due to the interference of various external environments such as space electromagnetic, the movement of the sensor, the difference in the response of the front and back of the detection unit, etc., the working parameters and working efficiency of the array pixel of the focal plane device can be affected, resulting in the output of the device. signal inconsistency

[0008] (5) Incompleteness and incorrectness of CCD response function correction

[0014] (1) Existing studies are all aimed at band noise with certain rules

But at present, the band noise of some satellites appears randomly, not every satellite image exists, and there is no obvious pattern

[0015] (2) Existing research mainly focuses on the research of noise removal methods, and the location judgment and noise type analysis of noise are mainly done manually, and it is difficult to realize fully automatic identification

Images