Image dislocation identification and correction method for infrared area array detector

Abstract

The invention discloses an image dislocation identification and correction method for an infrared area array detector, and solves the problem of space dimension and spectrum dimension errors caused by direct segmentation and splicing of dislocation images collected by the infrared area array detector. And difference calculation is carried out on the spatial position and a dislocation difference threshold value is set, so that the dislocation spatial position can be effectively identified. In combination with the characteristic of multi-channel reading of the area array detector, interval division is carried out on the original moon hyperspectral data in the spatial dimension, batch dislocation correction is carried out on the spectral dimension in each interval, the complexity and workload of dislocation pixel calculation are effectively reduced, and the image dislocation recognition and correction efficiency is improved.

Description

Technical Field

[0001] The invention relates to an infrared surface array detector image, and in particular to an infrared surface array detector image misalignment recognition and correction method. Background Art

[0002] Hyperspectral Remote Sensing (HRS) is a remote sensing technology that can obtain image data of dozens or even hundreds of continuous spectral bands in the ultraviolet to long-wave infrared region. Hyperspectral remote sensing data has the characteristics of image-spectrum integration, that is, each spatial pixel corresponds to a spectral curve, and each spectral band corresponds to a two-dimensional spatial image. Therefore, hyperspectral remote sensing data can provide rich spectral information and spatial information at the same time, and has the characteristics of large amount of information and high data transmission rate.

[0003] When using an infrared array detector for ground-based lunar shortwave infrared hyperspectral observation, the imaging spectrometer's slit field of view is pushed from the right edge of the moon to the left edge to complete a push-scan observation, and this operation is repeated to achieve long-term continuous observation of the moon. Due to temperature changes, environmental interference and other reasons, the sensitivity and working state of the detector may be affected. When the detector's collected data and output data are inconsistent, it will lead to the loss of collected data, which will cause the misalignment of the detector's output data.

[0004] Since the two-dimensional data collected by the detector in a single time is spectral dimension × spatial dimension data, it is difficult to directly judge whether the output is misaligned with a single frame of data. Taking advantage of the characteristics of the unified hyperspectral data spectrum, the data corresponding to a specific band can be selected from the single lunar observation data cube to obtain a two-dimensional spatial image. Direct segmentation and splicing may still have pixel-level misalignment, and cannot effectively correct the image errors in the spatial and spectral dimensions. If manual methods are used to identify and correct misalignment, it will be time-consuming and laborious due to the huge amount of data in hyperspectral remote sensing images. Therefore, high-performance, fast, and efficient processing methods are urgently needed to accurately identify and correct misalignment in hyperspectral data. Summary of the invention

[0005] In order to solve the technical problems that direct segmentation and splicing of misaligned images collected by infrared surface array detectors will result in spatial and spectral errors, or that manual identification and correction of misalignment will be time-consuming and laborious due to the huge amount of data of hyperspectral remote sensing images, the present invention provides a method for identifying and correcting misalignment of infrared surface array detector images.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A method for identifying image misalignment of an infrared array detector, which is special in that it comprises the following steps:

[0008] Step 1, obtaining the original lunar hyperspectral data D(f, p, b) collected by the array detector scanning the moon; and obtaining the array size P×B of the array detector, where P is the number of pixels in the spatial dimension of the detector array, B is the total number of spectral bands of the detector array, f=1,2,3...F, F is the total number of frames, p is the spatial position, p=1,2,3...P, b is the number of spectral bands, b=1,2,3...B;

[0009] Step 2: Extract the single-band spatial and temporal data of the original lunar hyperspectral data D(f, p, b) to obtain a single-band lunar image D b0 (f, p) = D(f, p, b0); where b0 is the number of bands without atmospheric absorption;

[0010] Step 3: Calculate the single-band lunar image D b0 The total difference S between the spatial positions p and p+1 in (f, p) in different frames p ;

[0011] Step 4: Obtain the misalignment difference threshold T;

[0012] Step 5: traverse the spatial position p = 1, 2, 3 ... P-1, and convert the difference S p The spatial position greater than the misalignment difference threshold T is recorded as the misalignment position pi, and the misalignment position pi is stored in the position index set R, and 0 and p=P are also stored in the position index set R;

[0013] Step 6: Sort the index values ​​in the position index set R from small to large, and calculate the difference between adjacent index values ​​to obtain the misaligned position interval number set RR;

[0014] Step 7, obtain the first value that is not P1 in the misaligned position interval number set RR, record it as the number of misaligned pixels pp, obtain the misalignment recognition result, and complete the misalignment recognition; P1=P / N, N is the number of readout modules of the array detector.

[0015] Furthermore, step 3 is specifically as follows:

[0016] Starting from the p=1th row spatial position, calculate the single-band moon image D line by line b0 The total difference S between each spatial position p and p+1 in (f, p) in different frames p :

[0017]

[0018] Furthermore, step 4 is specifically as follows:

[0019] Get the misalignment difference threshold T:

[0020]

[0021] in, For all differences S p The average value, k is the expansion multiple, 8≤k≤12.

[0022] Furthermore, in step 4, k=10.

[0023] A method for correcting image misalignment of an infrared array detector is characterized in that it comprises the following steps:

[0024] Step 1, obtaining the original lunar hyperspectral data D(f, p, b) collected by the array detector scanning the moon; and obtaining the array size P×B of the array detector, where P is the number of pixels in the spatial dimension of the detector array, B is the total number of spectral bands of the detector array, f=1,2,3...F, F is the total number of frames, p is the spatial position, p=1,2,3...P, b is the number of spectral bands, b=1,2,3...B;

[0025] Step 2: Extract the single-band spatial and temporal data of the original lunar hyperspectral data D(f, p, b) to obtain a single-band lunar image D b0 (f, p) = D(f, p, b0); where b0 is the number of bands without atmospheric absorption;

[0026] Step 3: Calculate the single-band lunar image D b0 The total difference S between the spatial positions p and p+1 in (f, p) in different frames p ;

[0027] Step 4: Obtain the misalignment difference threshold T;

[0028] Step 5, traverse the spatial position p = 1, 2, 3 ... P-1, and convert the difference S p The spatial position greater than the misalignment difference threshold T is recorded as the misalignment position p i , the dislocation position p i Store it in the position index set R, and also store 0 and p=P in the position index set R;

[0029] Step 6: Sort the index values ​​in the position index set R from small to large, and calculate the difference between adjacent index values ​​to obtain the misaligned position interval number set RR;

[0030] Step 7, obtaining the first value that is not P1 in the offset position interval number set RR, recorded as the offset pixel number pp; P1 = P / N, N is the number of readout modules of the array detector;

[0031] Step 8: Extract the spatial and spectral data of a single frame from the original lunar hyperspectral data D(f, p, b) to obtain a single-frame lunar image D f (p, b);

[0032] Step 9: For each frame of the moon image D f (p, b) is segmented in the spatial dimension with P1 as the interval, and each frame of the moon image D is obtained. f The sub-image set DD(n) to be corrected of (p, b), where n is the sub-image number, n=1, ... N;

[0033] Step 10: For each frame of the moon image D f Each sub-image DD(n) to be corrected of (p, b) is subjected to the following steps to obtain a corrected sub-image DD′(n) of each sub-image DD(n) to be corrected:

[0034] Step 10.1: Transform each frame of the moon image D f Each sub-image DD(n) to be corrected of (p, b) is rearranged along the spatial dimension direction to obtain a one-dimensional array DD1 corresponding to each sub-image DD(n) to be corrected and having a length of L=B×P1;

[0035] Step 10.2, delete the first pp data in each one-dimensional array DD1, and delete the last P1-pp data in each one-dimensional array DD1, to obtain a one-dimensional array DD1′;

[0036] Step 10.3, rearrange each one-dimensional array DD1′ along the spatial dimension with P1 as the interval, and obtain the corrected sub-image DD′(n) of each sub-image to be corrected DD(n), whose image size is (B-1)×P1;

[0037] Step 11: For each frame of the moon image D f (p, b), and concatenate the corresponding corrected sub-images DD′(n) in the spatial dimension according to the sub-image sequence number n to obtain each frame of the lunar image D f (p, b) The corrected single-frame lunar image D f ′(p, b), the image size is (B-1)×P;

[0038] Step 12: All single-frame moon images D f ′(p, b) are spliced ​​in the time dimension to obtain the hyperspectral data D′(f, p, b) after misalignment correction, completing the image misalignment correction.

[0039] Furthermore, step 3 is specifically as follows:

[0040] Starting from the p=1th row spatial position, calculate the single-band moon image D line by line b0 The total difference S between each spatial position p and p+1 in (f, p) in different frames p :

[0041]

[0042] Furthermore, step 4 is specifically as follows:

[0043] Get the misalignment difference threshold T:

[0044]

[0045] in, For all differences S p The average value, k is the expansion multiple, 8≤k≤12.

[0046] Furthermore, in step 4, k=10.

[0047] Furthermore, step 9 is specifically as follows:

[0048] Starting from the f=1 frame, each frame of the moon image D f (p, b) is segmented in the spatial dimension with P1 as the interval, and each frame of the moon image D is obtained. f The sub-image set DD(n) to be corrected of (p, b):

[0049] DD(n)=D f (p n , b)

[0050] Among them, p n =(n-1)*P1+1, (n-1)*P1+2, (n-1)*P1+3,...(n-1)*P1+P1.

[0051] Furthermore, in step 10, for each frame of the moon image D f Each sub-image DD(n) to be corrected of (p, b), starting from the n=1th sub-image DD(n) to be corrected, is corrected for each frame of the lunar image D f Each sub-image DD(n) to be corrected of (p, b) is corrected.

[0052] Beneficial effects of the present invention:

[0053] 1. The present invention provides an infrared array detector image misalignment recognition and correction method, which combines the characteristics of hyperspectral images in spatial continuity, performs difference calculation on spatial positions and sets misalignment difference thresholds, and can effectively identify misaligned spatial positions.

[0054] 2. The present invention provides an infrared surface array detector image misalignment recognition and correction method, which combines the characteristics of multi-channel readout of the surface array detector, divides the original lunar hyperspectral data into intervals in the spatial dimension, and performs batch misalignment correction on the spectral dimension within each interval, effectively reducing the complexity and workload of misaligned pixel calculation and improving the efficiency of image misalignment recognition and correction.

[0055] 3. The present invention provides an infrared array detector image misalignment recognition and correction method, which recognizes the misalignment of the misaligned image composed of spatial dimension and time dimension, and performs image segmentation and reordering in spatial dimension and spectral dimension, which effectively improves the correction accuracy of the original lunar hyperspectral data and is of great significance to the research on hyperspectral data processing technology.

[0056] 4. The infrared array detector image misalignment recognition and correction method provided by the present invention has great application prospects in the fields of astronomical observation and ground object recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a flow chart of an embodiment of a method for identifying and correcting image misalignment of an infrared array detector according to the present invention;

[0058] Figure 2 is a schematic diagram of an original lunar hyperspectral data cube in an embodiment of the present invention;

[0059] Figure 3 3 is a comparison diagram before and after data processing in an embodiment of the present invention; wherein (a) is the original single-band image of the moon, the green dotted line is the boundary line of the readout module, and the red solid line is the identified misaligned spatial position; (b) is the corrected single-band image of the moon. DETAILED DESCRIPTION

[0060] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] The infrared array detector image misalignment recognition method provided by the embodiment of the present invention is combined with Figure 1 and Figure 2 As shown, the following steps are included:

[0062] Step 1, obtain the original lunar hyperspectral data D (f, p, b) collected by the array detector scanning the moon; and obtain the array size P×B of the array detector, where P is the number of pixels in the spatial dimension of the detector array, B is the total number of spectral bands of the detector array, f=1,2,3...F, F is the total number of frames, p is the spatial position, p=1,2,3...P, b is the number of spectral bands, b=1,2,3...B;

[0063] Step 2: Obtain single-band lunar image D b0 (f, p);

[0064] The original lunar hyperspectral data D(f, p, b) is extracted from the single-band spatial dimension and time dimension to obtain the single-band lunar image D b0 (f, p):

[0065] D b0 (f, p) = D (f, p, b0)

[0066] Among them, b0 is the number of bands without atmospheric absorption;

[0067] Step 3: Initialize the spatial position p, set p = 1;

[0068] Starting from the p=1th row of spatial positions, calculate the overall difference S between each spatial position p and p+1 in different frames row by row p :

[0069]

[0070] Make sure every spatial position participates in the calculation;

[0071] Step 4: Get the misalignment difference threshold T:

[0072]

[0073] in, For all differences S p The average value of k is the expansion multiple, 8≤k≤12; in this embodiment, k=10, that is

[0074] Step 5: Traverse the spatial position p = 1, 2, 3 ... P-1, and convert each difference S p Compare with the misalignment difference threshold T; S p Greater than T corresponding to the spatial position p i Recorded as the dislocation position, the dislocation position p i Store it in the position index set R, and also store 0 and p=P in the position index set R;

[0075] Step 6: Sort the index values ​​in the position index set R from small to large, and calculate the difference between adjacent index values ​​to obtain the misaligned position interval number set RR;

[0076] Step 7: Get the first value that is not P1 in the offset position interval number set RR, and record it as the offset pixel number pp, such as Figure 3 As shown in (a), the misalignment recognition result is obtained and the misalignment recognition is completed; P1 = P / N, N is the number of readout modules of the array detector.

[0077] An infrared array detector image misalignment correction method provided by the embodiment of the present invention is combined with Figure 1 and Figure 2As shown, the following steps are included:

[0078] Step 1, obtain the original lunar hyperspectral data D (f, p, b) collected by the array detector scanning the moon; and obtain the array size P×B of the array detector, where P is the number of pixels in the spatial dimension of the detector array, B is the total number of spectral bands of the detector array, f=1,2,3...F, F is the total number of frames, p is the spatial position, p=1,2,3...P, b is the number of spectral bands, b=1,2,3...B;

[0079] Step 2: Extract the single-band spatial and temporal data of the original lunar hyperspectral data D(f, p, b) to obtain a single-band lunar image D b0 (f, p):

[0080] D b0 (f, p) = D (f, p, b0)

[0081] Among them, b0 is the number of bands without atmospheric absorption;

[0082] Step 3: Starting from the p=1th row of spatial positions, calculate the overall difference S between each spatial position p and p+1 in different frames row by row p :

[0083]

[0084] Make sure every spatial position participates in the calculation;

[0085] Step 4: Get the misalignment difference threshold T:

[0086]

[0087] in, For all differences S p The average value of k is the expansion multiple, 8≤k≤12; in this embodiment, k=10 is preferred.

[0088] Step 5: Traverse the spatial position p = 1, 2, 3 ... P-1, and convert each difference S p Compare with the misalignment difference threshold T; S p Greater than T corresponding to the spatial position p i Recorded as the dislocation position, the dislocation position p i Store it in the position index set R, and also store 0 and p=P in the position index set R;

[0089] Step 6: Sort the index values ​​in the position index set R from small to large, and calculate the difference between adjacent index values ​​to obtain the misaligned position interval number set RR;

[0090] Step 7, obtain the first value that is not P1 in the offset position interval number set RR, and record it as the offset pixel number pp; P1 = P / N, N is the number of readout modules of the array detector;

[0091] Step 8: Extract the spatial and spectral data of a single frame from the original lunar hyperspectral data D(f, p, b) to obtain a single-frame lunar image D f (p, b);

[0092] Step 9: Initialize the frame number f, set f = 1, and start from the first frame to perform the lunar image D f (p, b) is segmented in the spatial dimension with P1 as the interval, and each frame of the moon image D is obtained. f The sub-image set DD(n) to be corrected of (p, b):

[0093] DD(n)=D f (p n , b)

[0094] Where n is the sub-graph number, n = 1, ... N, p n =(n-1)*P1+1, (n-1)*P1+2, (n-1)*P1+3,...(n-1)*P1+P1;

[0095] Step 10: For each frame of the moon image D f For each sub-image DD(n) to be corrected of (p, b), the sub-image number n is initialized, and n=1. Starting from the first sub-image DD(n) to be corrected, each frame of the lunar image D f Each sub-image DD(n) to be corrected of (p, b) is subjected to the following steps to obtain a corrected sub-image DD′(n) of each sub-image DD(n) to be corrected:

[0096] Step 10.1: Transform each frame of the moon image D f Each sub-image DD(n) to be corrected of (p, b) is rearranged along the spatial dimension direction to obtain a one-dimensional array DD1 corresponding to each sub-image DD(n) to be corrected and having a length of L=B×P1;

[0097] Step 10.2, delete the first pp data in each one-dimensional array DD1, and delete the last P1-pp data in each one-dimensional array DD1, to obtain a one-dimensional array DD1′;

[0098] Step 10.3, rearrange each one-dimensional array DD1′ along the spatial dimension with P1 as the interval, and obtain the corrected sub-image DD′(n) of each sub-image to be corrected DD(n), whose image size is (B-1)×P1;

[0099] Step 11: For each frame of the moon image Df (p, b), and concatenate the corresponding corrected sub-images DD′(n) in the spatial dimension according to the sub-image sequence number n to obtain each frame of the lunar image D f (p, b) The corrected single-frame lunar image D f ′(p, b), the image size is (B-1)×P;

[0100] Step 12: All single-frame moon images D f ′(p, b) are spliced ​​in the time dimension to obtain the hyperspectral data D′(f, p, b) after misalignment correction, completing the image misalignment correction.

[0101] The single-band spatial and temporal data of the misaligned hyperspectral data D′(f, p, b) are extracted to obtain Figure 3 (b) shows a single-band lunar image. It can be seen that the recognition and correction methods provided in this embodiment have excellent effects.

[0102] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for identifying image misalignment of an infrared array detector, characterized in that: The following steps are involved: Step 1, obtaining the original lunar hyperspectral data D (f, p, b) collected by the array detector scanning the moon; and obtaining the array size P×B of the array detector, where P is the number of pixels in the spatial dimension of the detector array, B is the total number of spectral bands of the detector array, f=1,2,3...F, F is the total number of frames, p is the spatial position, p=1,2,3...P, b is the number of spectral bands, b=1,2,3...B; Step 2: Extract the single-band spatial and temporal data of the original lunar hyperspectral data D(f, p, b) to obtain a single-band lunar image D b0 (f, p) = D(f, p, b0); where b0 is the number of bands without atmospheric absorption; Step 3: Calculate the single-band lunar image D b0 The total difference S between the spatial positions p and p+1 in (f, p) in different frames p ; Step 4: Obtain the misalignment difference threshold T; Step 5: traverse the spatial position p=1,2,3...P-1, and convert the difference S p The spatial position greater than the misalignment difference threshold T is recorded as the misalignment position pi, and the misalignment position pi is stored in the position index set R, and 0 and p=P are also stored in the position index set R; Step 6: Sort the index values ​​in the position index set R from small to large, and calculate the difference between adjacent index values ​​to obtain the misaligned position interval number set RR; Step 7, obtain the first value that is not P1 in the misaligned position interval number set RR, record it as the number of misaligned pixels pp, obtain the misalignment recognition result, and complete the misalignment recognition; P1=P / N, N is the number of readout modules of the array detector.

2. The infrared array detector image misalignment recognition method according to claim 1, characterized in that: Step 3 is as follows: Starting from the p=1th row spatial position, calculate the single-band moon image D line by line b0 The total difference S between each spatial position p and p+1 in (f, p) in different frames p :

3. The infrared array detector image misalignment recognition method according to claim 2, characterized in that: Step 4 is as follows: Get the misalignment difference threshold T: in, For all differences S p The average value, k is the expansion multiple, 8≤k≤12.

4. The infrared array detector image misalignment recognition method according to claim 3, characterized in that: In step 4, k=10.

5. A method for correcting image misalignment of an infrared array detector, characterized in that: The following steps are involved: Step 1, obtaining the original lunar hyperspectral data D (f, p, b) collected by the array detector scanning the moon; and obtaining the array size P×B of the array detector, where P is the number of pixels in the spatial dimension of the detector array, B is the total number of spectral bands of the detector array, f=1,2,3...F, F is the total number of frames, p is the spatial position, p=1,2,3...P, b is the number of spectral bands, b=1,2,3...B; Step 2: Extract the single-band spatial and temporal data of the original lunar hyperspectral data D(f, p, b) to obtain a single-band lunar image D b0 (f, p) = D(f, p, b0); where b0 is the number of bands without atmospheric absorption; Step 3: Calculate the single-band lunar image D b0 The total difference S between the spatial positions p and p+1 in (f, p) in different frames p ; Step 4: Obtain the misalignment difference threshold T; Step 5, traverse the spatial position p = 1, 2, 3 ... P-1, and convert the difference S p The spatial position greater than the misalignment difference threshold T is recorded as the misalignment position p i , the dislocation position p i Store it in the position index set R, and also store 0 and p=P in the position index set R; Step 6: Sort the index values ​​in the position index set R from small to large, and calculate the difference between adjacent index values ​​to obtain the misaligned position interval number set RR; Step 7, obtaining the first value that is not P1 in the offset position interval number set RR, recorded as the offset pixel number pp; P1 = P / N, N is the number of readout modules of the array detector; Step 8: Extract the spatial and spectral data of a single frame from the original lunar hyperspectral data D(f, p, b) to obtain a single-frame lunar image D f (p, b); Step 9: For each frame of the moon image D f (p, b) is segmented in the spatial dimension with P1 as the interval, and each frame of the moon image D is obtained. f The sub-image set DD(n) to be corrected of (p, b), where n is the sub-image number, n=1, ... N; Step 10: For each frame of the moon image D f Each sub-image DD(n) to be corrected of (p, b) is subjected to the following steps to obtain a corrected sub-image DD′(n) of each sub-image DD(n) to be corrected: Step 10.1: Transform each frame of the moon image D f Each sub-image DD(n) to be corrected of (p, b) is rearranged along the spatial dimension direction to obtain a one-dimensional array DD1 corresponding to each sub-image DD(n) to be corrected and having a length of L=B×P1; Step 10.2, delete the first pp data in each one-dimensional array DD1, and delete the last P1-pp data in each one-dimensional array DD1, to obtain a one-dimensional array DD1′; Step 10.3, rearrange each one-dimensional array DD1′ along the spatial dimension with P1 as the interval, and obtain the corrected sub-image DD′(n) of each sub-image to be corrected DD(n), whose image size is (B-1)×P1; Step 11: For each frame of the moon image D f (p, b), and concatenate the corresponding corrected sub-images DD′(n) in the spatial dimension according to the sub-image sequence number n to obtain each frame of the lunar image D f (p, b) The corrected single-frame lunar image D f ′(p, b), the image size is (B-1)×P; Step 12: All single-frame moon images D f ′(p, b) are spliced ​​in the time dimension to obtain the hyperspectral data D′(f, p, b) after misalignment correction, completing the image misalignment correction.

6. The infrared array detector image misalignment correction method according to claim 5, characterized in that: Step 3 is as follows: Starting from the p=1th row spatial position, calculate the single-band moon image D line by line b0 The total difference S between each spatial position p and p+1 in (f, p) in different frames p :

7. The infrared array detector image misalignment correction method according to claim 6, characterized in that: Step 4 is as follows: Get the misalignment difference threshold T: in, For all differences S p The average value, k is the expansion multiple, 8≤k≤12.

8. The method for correcting the image misalignment of an infrared array detector according to claim 7, characterized in that: In step 4, k=10.

9. The infrared array detector image misalignment correction method according to any one of claims 5 to 8, characterized in that: Step 9 is as follows: Starting from the f=1 frame, each frame of the moon image D f (p, b) is segmented in the spatial dimension with P1 as the interval, and each frame of the moon image D is obtained. f The sub-image set DD(n) to be corrected of (p, b): DD(n)=D f (p n ,b) Among them, p n =(n-1)*P1+1, (n-1)*P1+2, (n-1)*P1+3,...(n-1)*P1+P1.

10. The infrared array detector image misalignment correction method according to claim 9, characterized in that: In step 10, for each frame of the moon image D f Each sub-image DD(n) to be corrected of (p, b) starts from the n=1th sub-image DD(n) to be corrected, and each frame of the lunar image D f Each sub-image DD(n) to be corrected of (p, b) is corrected.

Images