An Optical Payload Blind Pixel Discrimination Method Based on Yaw Data

Abstract

The present invention discloses a method for discriminating blind pixels of an optical payload based on yaw data, belonging to the technical field of remote sensing data processing, including: (1) obtaining yaw data in the optical payload; (2) preprocessing the yaw data; (3) performing inter-frame difference on the yaw data processed in step two to obtain a difference image; taking the average value of each frame of the difference image to obtain a difference average image, and the gray value corresponding to each pixel in the difference average image is denoted as DN i,j , taking the average value and standard deviation of each band of pixels in one row of the difference average image (4) The method for discriminating blind pixels of the optical payload based on yaw data is: when η is a natural number not less than 3 (setting corresponding thresholds according to the characteristics of each sensor) or DN i,j = 0, it is determined as a blind pixel. This method analyzes the data to find the characteristics of blind pixels different from other pixels, thereby setting the determination conditions to quickly, effectively, and accurately determine the positions of blind pixels.

Description

Technical Field

[0001] The present invention relates to the technical field of hyperspectral data processing, and particularly to a method for discriminating blind pixels of an optical payload based on yaw data. Background Art

[0002] Infrared imaging systems are increasingly widely used, and the requirements for imaging quality are getting higher and higher. However, the problem of blind pixels inevitably exists in infrared focal plane array imaging, which greatly reduces the spatial resolution and temperature resolution of the imaging system, seriously restricts the performance of the imaging system, and extremely affects the imaging quality. Therefore, it is necessary to solve the problem of blind pixels. As a prerequisite for blind pixel repair of imaging quality, blind pixel detection technology has become increasingly important.

[0003] Currently, there are three classic blind pixel detection algorithms: the national standard method, the double-reference source difference detection algorithm, and the blind pixel detection algorithm based on the 3σ principle. These algorithms have the disadvantages of cumbersome steps, fixed dead pixels in detection, long time, the influence of the test area on accuracy, and poor applicability after being on orbit because these methods are all for the laboratory stage. The uneven response bias, inherent noise, inconsistent dark current, and inconsistent peripheral circuits of each pixel of the optical detector will all lead to differences in the response between pixels. Therefore, the national standard clearly defines blind pixels in optical detectors: 1. Fixed blind pixels are pixels that always remain in a saturated state or a cut-off state and pixels with a constant gray value. 2. Random blind pixels are pixels with high noise and have nothing to do with the environment.

[0004] In the yaw imaging mode, all pixels of the CCD will image the same ground object in sequence. After normalization, each row of the yaw data corresponds to the same ground object, and the response should be consistent. If the response is too abnormal, it can be determined as a blind pixel. Using the yaw data to discriminate blind pixels can greatly avoid the influence of factors such as ground object differences and noise on the determination result and improve the determination accuracy. Summary of the Invention

[0005] The embodiments of the present invention provide a method for discriminating blind pixels of an optical payload based on yaw data, which can use the yaw data to determine the positions of blind pixels of the corresponding payload, achieving the advantages of fast in-orbit blind pixel search, low false detection rate of blind pixels, and high efficiency. The content of the present invention is as follows:

[0006] The purpose of the present invention is to provide a method for discriminating blind pixels of an optical payload based on yaw data. The technical points are as follows: including the following steps:

[0007] Step 1, obtain N frames of yaw data of the optical payload;

[0008] Step 2, preprocess the yaw data obtained in Step 1 to obtain the normalized yaw data, and obtain the gray value corresponding to each pixel according to the normalized yaw data;

[0009] Step 3: Perform inter-frame difference on the normalized yaw data obtained in Step 2 to obtain (N - 1) frames of yaw data. Take the average value of the yaw data after difference for each band to obtain a difference mean image, and record the gray value corresponding to each pixel in the difference mean image as DN i,j , and take the average value of the difference mean image for each band and standard deviation

[0010] Step 4: The method for discriminating blind pixels of the optical payload based on yaw data is as follows: When where η is a natural number not less than 3 or DN i,j = 0, it is discriminated as a blind pixel; otherwise, it is a non-blind pixel.

[0011] In some embodiments of the present invention, in the method for discriminating blind pixels of the optical payload based on yaw data of the present invention, the yaw data obtained in Step 1 is the yaw data of a spaceborne hyperspectral satellite.

[0012] In some embodiments of the present invention, in the method for discriminating blind pixels of the optical payload based on yaw data of the present invention, the preprocessing methods of the yaw data in Step 2 include data separation processing, data dark level processing, and data normalization processing.

[0013] In some embodiments of the present invention, the specific method of yaw data separation processing in the method for discriminating blind pixels of the optical payload based on yaw data of the present invention is as follows: The yaw data in Step 1 is a binary file, including image data, seeker data, and auxiliary data. The required image data and frame number are separated from the binary file through the data storage format.

[0014] In some embodiments of the present invention, the specific method of yaw data dark level processing in the method for discriminating blind pixels of the optical payload based on yaw data of the present invention is as follows: Establish the relationship between the frame number and the dark level gray value through on-orbit dark level data. Calculate the corresponding dark level coefficient according to the frame number of the separated yaw data. Then subtract the dark level coefficient from the separated yaw image data to obtain the yaw data after deducting the dark level.

[0015] In some embodiments of the present invention, the specific operation of yaw data normalization processing in the method for discriminating blind pixels of the optical payload based on yaw data of the present invention is as follows: Use the line segment detection method to detect the yaw angle, and then normalize and align the data after deducting the dark level. For the normalized data, each pixel in each row images the same ground object.

[0016] The above at least one technical solution adopted in the embodiments of the present invention can achieve the following beneficial effects:

[0017] The method for discriminating blind pixels of an optical payload based on yaw data of the present invention includes: (1) obtaining yaw data of N frames of the optical payload; (2) preprocessing the yaw data obtained in step (1) to obtain standardized yaw data; (3) performing inter-frame difference on the new yaw data obtained in step (2) to obtain (N - 1) frames of yaw data, taking the average value of the differential yaw data by band to obtain a differential mean image, and the gray value corresponding to each pixel in the differential mean image is denoted as DN i,j taking the average value of the differential mean image by band and standard deviation (4) The method for discriminating blind pixels of the optical payload based on yaw data is: when where η is a natural number not less than 3 or DN i,j = 0, it is discriminated as a blind pixel, otherwise it is a non-blind pixel. This method analyzes the data to find the characteristics of blind pixels different from other pixels, thereby setting the determination conditions to quickly, effectively and accurately determine the position of blind pixels. Specific embodiments

[0018] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with specific embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0019] A method for discriminating blind pixels of an optical payload based on yaw data includes the following steps:

[0020] Step 1: Obtain yaw data of N frames of a hyperspectral satellite. In this example, the on-orbit hyperspectral yaw data of short-wave yaw data is used.

[0021] Step 2: Preprocess the yaw data obtained in step 1 to obtain new yaw data;

[0022] Preferably, in the method for discriminating blind pixels of the optical payload based on yaw data of the present invention, the preprocessing method of the yaw data in step 2 includes data separation processing, data dark level processing, and data standardization processing. Specifically, data separation is performed on the format of the on-orbit hyperspectral yaw data, dark level deduction is performed according to the on-orbit dark level data, the deviation angle is calculated using a line segment detection (such as LSD) method, and then the data after dark level deduction is standardized. Each pixel in each row of the standardized data of the present invention images the same ground object, while each pixel in each row of the yaw data images different ground objects. When setting the threshold, the influence of errors caused by ground object inconsistency needs to be considered, while the yaw data does not need to be considered.

[0023] Further, in the method for discriminating blind pixels of an optical payload based on yaw data according to the present invention, the specific method for separating and processing yaw data is as follows: the yaw data in step one is a binary file, and the required image data is separated through the data storage format.

[0024] Further, in the method for discriminating blind pixels of an optical payload based on yaw data according to the present invention, the specific method for processing the dark level of yaw data is as follows: by using the dark level data, the relationship between the frame number and the dark level gray value is established, and then the separated yaw image data is subtracted by the dark level data to obtain the dark level data deducted.

[0025] Further, the specific operation of the normalization processing of the yaw data in the method for discriminating blind pixels of an optical payload based on yaw data according to the present invention is as follows: the angle of the stripe is calculated by using the line segment detection method, and then the data after deducting the dark level is normalized and aligned. After normalization, each pixel in each row images the same ground object.

[0026] Step three, perform inter-frame difference on the new yaw data obtained in step two to obtain (N - 1) frames of yaw data, and take the average value of the differential yaw data by band to obtain a differential mean image. The gray value corresponding to each pixel in the differential mean image is denoted as DN i,j , take the average value of the differential mean image by band and standard deviation

[0027] Step four, the method for discriminating blind pixels of an optical payload based on yaw data is as follows:

[0028] When where η is a natural number not less than 3, it represents a pixel with too low or too high response or a flash pixel, that is, a pixel with an unfixed gray value and not equal to 0, presenting black lines, white lines, pixels presenting along-track flashing and discontinuity

[0029] Or,

[0030] DN i,j = 0, it means that the gray value corresponding to each pixel is 0, the gray value corresponding to each pixel is the saturation value, or each pixel corresponds to a fixed value.

[0031] If the above phenomena occur, it can be determined as a blind pixel, otherwise it is a non-blind pixel.

[0032] It meets the definition of blind pixels in the national standard.

[0033] Taking the blind pixels found visually as the standard, compare the blind pixels determined by the above method with the results of the blind pixels found by the visual method, and evaluate the blind pixel discrimination method of the present invention from the aspects of missed detection and false detection. The evaluation results are as follows:

[0034] (1) Missed detection: Through calculation and comparison, the missed detection of blind pixel discrimination by the above method is approximately 20 - 25 pixels (the total number of pixels is 2048 * 180). This missed detection rate is within a reasonable range. Through observation and comparison, it is found that half of the missed pixels are in the atmospheric absorption channel or spectral bands with high noise.

[0035] (2) False detection: Through calculation and comparison, the false detection of blind pixel discrimination by the above method is 300 pixels. Through observation and comparison, it is found that nearly 2 / 3 of the pixels are caused by noise, mainly in the atmospheric absorption channel or spectral bands with more noise. The greater the noise, the easier it is to be misdetected as a blind pixel because the mean and variance of this pixel will be higher than those of the surrounding pixels. For example, wavelengths 1354.082 - 1438.248, 1808.580 - 1960.079, 2372.494 - 2515.577 nm. Therefore, for spectral bands with strong noise, two methods can be provided. Either only pixels with DN i,j = 0 are regarded as blind pixels, or η is set higher to increase the threshold limit, which can greatly reduce the false detection rate and will not overly increase the missed detection rate.

[0036] The above are only embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, various changes and modifications can be made to the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the scope of the claims of the present invention.

Claims

1. An optical payload blind pixel discrimination method based on yaw data, characterized in that Including the following steps: Step 1: Obtain N frames of yaw data with optical payloads; Step 2: Preprocess the yaw data obtained in Step 1 to obtain new yaw data. The preprocessing methods of the yaw data include data separation processing, data dark level processing, and data normalization processing. The specific operation of the yaw data normalization processing is as follows: Use the line segment detection method to detect the yaw angle, and then normalize and align the data after subtracting the dark level. For the normalized data, each pixel in each row images the same ground object; Step 3: Perform inter-frame difference on the new yaw data obtained in Step 2 to obtain (N - 1) differential images. Calculate the average value of the yaw data after difference for each band to obtain a differential mean image, and record the gray value corresponding to each pixel in the differential mean image as DN i,j , calculate the average value of the differential mean image for each band and standard deviation Step 4, the method for discriminating blind pixels of the optical payload based on yaw data is: when where η is a natural number not less than 3, or, DN i,j = 0, it is discriminated as a blind pixel, otherwise it is a non-blind pixel.

2. The method for discriminating blind pixels of an optical payload based on yaw data according to claim 1, wherein, The specific method of the yaw data separation processing is as follows: The yaw data in Step 1 is a binary file, including image data, seeker data, and auxiliary data. The binary file separates the required image data and frame number through the data storage format.

3. A method for discriminating blind pixels of an optical payload based on yaw data according to claim 1, characterized in that, The specific method of the yaw data dark level processing is as follows: Establish the relationship between the frame number and the dark level gray value through the on-orbit dark level data. According to the frame number of the separated yaw data, calculate the corresponding dark level coefficient, and then subtract the dark level coefficient from the separated yaw image data to obtain the yaw data after subtracting the dark level.

Images