CN104601906A - Automatic suppression method for on-orbit noise of image detector - Google Patents

Abstract

The present invention relates to an image detector on-orbit automatic noise suppression method, comprising the following steps: 1] starting to read a new frame of image output by the image detector; 2) performing target detection on the output image; 3] performing the entire frame All pixels of the image are processed by IIR adaptive filtering, and the noise correction amount of pixel grayscale is calculated and stored; 4] The original grayscale of all pixels in the entire frame of image is compared with the noise correction amount of corresponding pixel grayscale to obtain The corrected gray scale of the whole frame image. The present invention solves the technical problems of large amount of calculation, waste of storage resources, and the need to continuously upload calibration data in the prior art for on-board software to remove white noise points, dark current non-uniformity, and fixed pattern noise. The present invention can maximize real-time Remove the effects of fixed pattern noise, dark current non-uniformity and white noise.

Description

On-orbit noise automatic suppression method for image detector

technical field

The invention relates to an image detector on-orbit automatic noise suppression method, which can realize automatic compensation of the image detector noise on the satellite.

Background technique

The image detector, also called the image sensor, is the core part of the spaceborne camera. According to the different components, it can be divided into two categories: CCD (Charge Coupled Device, charge coupled device) and CMOS (Complementary Metal-Oxide Semiconductor, metal oxide semiconductor component).

Although CCD and CMOS are quite different in technology, the performance gap between CCD and CMOS is not very large. At present, CCD and CMOS image detectors are widely used in fields such as astronomical observation, visible light imaging and space optics.

The output signals of CCD and CMOS image detectors are mixed with various noises and interferences, mainly including thermal noise, dark current noise, shot noise, reset noise and noise introduced by the readout circuit. These noises will have a serious impact on image quality, manifested as white noise, dark current non-uniformity and fixed pattern noise on the image. Although these noises are suppressed in the hardware circuit, the influence of these noises cannot be completely eliminated, and on-board software processing is generally required.

The method of processing the white noise in the image is to store the position information of the white noise in the memory in advance, and when the image detector outputs to the noise, the grayscale of the previous or next pixel is used as the grayscale output of the noise .

The processing method for the non-uniformity of dark current is to store the correction coefficient of each pixel, and when the image detector outputs the gray level of the pixel, multiply each pixel by the corresponding correction coefficient and then output it.

Since the fixed pattern noise unique to the CMOS image detector is related to temperature, calibration images at different temperatures need to be stored, and calibration images at corresponding temperatures need to be subtracted from each frame of image.

The above-mentioned on-board software processing method needs to consume a lot of satellite storage resources and computing resources; when the performance of the image sensor in orbit degrades, the calibration data needs to be uploaded again; but this can only be maintained for a period of time. Further performance degradation requires continuous uploading of calibration data.

Contents of the invention

In order to solve the technical problems of large amount of calculation, waste of storage resources, and the need to upload calibration data continuously in the prior art for on-board software to remove white noise points, dark current non-uniformity, and fixed pattern noise, the present invention provides an image detector in Automatic rail noise suppression method.

Technical solution of the present invention:

An image detector on-orbit automatic noise suppression method is special in that it includes the following steps:

1] Start to read a new frame of image output by the image detector;

2] Perform target detection on the output image:

If no target is detected, proceed to step 3]; if one or more targets are detected, proceed to step 4];

3] Perform IIR adaptive filter processing on all pixels of the entire frame image, calculate and store the noise correction amount of the pixel grayscale, and go to step 5];

4] Perform IIR adaptive filter processing on all pixels except the area occupied by the target, calculate and store the noise correction amount of the pixel grayscale; the noise correction amount of the pixel grayscale in the target area is taken from the corresponding value of the previous frame image The correction amount of the pixel gray level of the area, and store it;

5] Make the difference between the original grayscale of all pixels in the entire frame image and the noise correction amount of the corresponding pixel grayscale to obtain the corrected entire frame image grayscale, and return to step 1].

The advantages that the present invention has:

1. The present invention can remove the influence of fixed pattern noise, dark current inhomogeneity and white noise to the greatest extent in real time.

2. The present invention minimizes the waste of on-board storage resources, and only needs one frame of image storage resources.

3. The present invention can self-adapt to the characteristic changes of fixed pattern noise, dark current non-uniformity and white noise point.

4. The present invention does not need to continuously upload calibration data on the ground, and has the advantage of autonomy.

5. The present invention can ensure that the spaceborne image detector works at relatively high temperature.

Description of drawings

Fig. 1 is a flowchart of the present invention;

Fig. 2 is the flowchart of IIR adaptive filtering;

Fig. 3 is a schematic diagram of an embodiment of the present invention.

Detailed ways

An image detector on-orbit automatic noise suppression method includes the following steps:

1] Start to read a new frame of image output by the image detector;

2] Perform target detection on the output image:

If no target is detected, proceed to step 3]; if one or more targets are detected, proceed to step 4];

3] Perform IIR adaptive filtering processing on all pixels of the entire frame image, calculate the fixed pattern noise correction amount of the pixel grayscale and store it, go to step 5]; the process of IIR adaptive filtering processing is shown in Figure 2, where S It is a smoothing factor, which can be 2, 4, 8, 16, 32.

4] Perform IIR adaptive filter processing on all pixels except the area occupied by one or more targets, calculate the fixed mode noise correction amount of the pixel gray level and store it, the pixels in the target area are not processed, and the noise of the corresponding pixels The correction amount is the same as the noise correction amount of the corresponding area in the previous frame;

5] Make a difference between the original grayscale of all pixels of the entire frame image and the corresponding correction amount to obtain the corrected entire frame image grayscale, and go to step 1].

Example:

An image detector on-orbit automatic noise suppression method includes the following steps:

1] Start to read a new frame of image output by the image detector;

2] Perform target detection on the output image:

If no target is detected, proceed to step 3]; if one or more targets are detected, proceed to step 4];

3] Perform IIR adaptive filtering processing on all pixels of the entire frame of image, calculate and store the fixed pattern noise correction amount of the pixel grayscale, go to step 5]; the process of IIR adaptive filtering processing is shown in Figure 2, where S It is a smoothing factor, which can be 2, 4, 8, 16, 32.

4] Perform IIR adaptive filter processing on all pixels except the area occupied by one or more targets, calculate the fixed mode noise correction amount of the pixel gray level and store it, the pixels in the target area are not processed, and the noise of the corresponding pixels The correction amount is the same as the noise correction amount of the corresponding area in the previous frame;

5] Make a difference between the original grayscale of all pixels of the entire frame image and the corresponding correction amount to obtain the corrected entire frame image grayscale, and go to step 1].

Example:

a: The image detector starts to work, and the target has not been detected on the image detector, and IIR adaptive filtering is performed on all pixels of the entire frame image, and the noise correction amount of the pixel gray scale is calculated;

b: When the target (such as a satellite) is detected on the image detector, the noise correction amount is not calculated for the pixels in the area occupied by the target, and the noise correction amount is calculated for the pixels outside the area occupied by the target;

c: The target moves to other positions on the target surface of the image detector, and the noise correction amount is not calculated for the pixels in the area occupied by the target, and the noise correction amount is calculated for the pixels outside the area occupied by the target;

d: The target moves out of the target surface of the image detector, that is, there is no target in the image detector, and the noise correction amount is calculated for all pixels of the entire frame image.

Claims (1)

1. the image detector automatic suppressing method of noise in-orbit, is characterized in that: comprise the following steps:

1] a new two field picture of the output of reading images detector is started;

2] target detection is carried out at the image exported:

If the target of not detecting, then proceed to step 3]; If one or more target detected, then proceed to step 4];

3] carry out the process of IIR adaptive-filtering to all pixels of whole two field picture, the noise correction amount of calculating pixel gray scale also stores, and goes to step 5];

4] carry out the process of IIR adaptive-filtering to except extra-regional all pixels shared by target, the noise correction amount of calculating pixel gray scale also stores; The noise correction of the pixel grey scale in target region measures the correction of the pixel grey scale of the respective regions of previous frame image, and stores;

5] original gradation of all pixels of whole two field picture and the noise correction amount of corresponding pixel grey scale are done difference, obtain revised whole two field picture gray scale, get back to step 1].