Method and device for manufacturing non-uniform data set of detector, electronic equipment and medium

Abstract

The invention provides a detector non-uniform data set manufacturing method and device, electronic equipment and a medium, and belongs to the field of detector imaging. The method comprises the following steps: acquiring a detector response image and a scene image; determining an ideal pixel value according to the detector response image, and preprocessing the scene image to obtain an original image; and constructing a data degradation model according to the ideal pixel value, performing data degradation processing on the original image according to the data degradation model to obtain a noise image, and constructing a detector non-uniform data set according to the noise image and the original image. According to the invention, data degradation processing is carried out on the high-quality scene image by using the data degradation model constructed by the detector response image, so that the high-quality detector non-uniform data set containing various non-uniform characteristics can be generated with high efficiency and low cost.

Description

Technical Field

[0001] The present invention relates to the technical field of detector imaging, and particularly to a method, device, electronic device and medium for producing a detector non-uniform data set. Background Art

[0002] In practical applications of detectors, due to the non-ideality of their sensitive elements and readout circuits, response non-uniformity problems often occur, that is, different detection units have different responses to the same radiation source. This non-uniformity will reduce the quality of the image and affect the accuracy of subsequent image processing and analysis. At present, detector non-uniformity correction algorithms are mainly divided into three categories: calibration-based correction algorithms, scene-based correction algorithms and deep learning algorithms. Calibration-based correction algorithms rely on periodic calibration of a reference radiation source, which is simple and easy to implement, but often cannot handle complex non-uniformity problems and have poor adaptability to environmental changes. Scene-based correction algorithms directly use scene information for correction without an additional calibration source, but they also have poor adaptability to environmental changes. Deep learning algorithms have been widely used in image non-uniformity correction due to their excellent performance in image processing and recognition tasks. With the development of deep learning technology, using a data set to train a deep learning model for non-uniformity correction has become a research hotspot.

[0003] However, in deep learning algorithms, the construction of the data set is the core issue. Since the non-uniformity differences shown by detectors under different environments and working conditions are relatively large, a data set containing various non-uniformity patterns is required. However, high-quality image data sets are difficult to obtain because they require specific equipment and environmental conditions; at the same time, due to the possible existence of blind pixels in the detector to be corrected, the collected images cannot restore the real details. To solve this problem, an existing method is to obtain detector images through laboratory environments or on-site acquisitions and construct a data set. However, this method is costly, the image acquisition is complex, and it is difficult to cover all possible non-uniformity situations. Another way is to use an existing data set as the original image and manually add vertical stripes or dot-like non-uniformity noise. Although this method has a lower cost, the simulated data set may not fully reflect the non-uniformity characteristics of actual detector images, affecting the accuracy of the algorithm. Therefore, existing data set production methods are difficult to produce a high-quality detector non-uniform data set efficiently and at low cost.

[0004] Therefore, there is a technical problem in the prior art that it is difficult to produce a high-quality detector non-uniform data set efficiently and at low cost, which needs to be improved. Summary of the Invention

[0005] In view of this, it is necessary to provide a method, apparatus, electronic device and medium for producing a detector non-uniform data set, which are used to solve the technical problem in the prior art that it is difficult to produce a high-quality detector non-uniform data set efficiently and at low cost.

[0006] To solve the above problems, on the one hand, the present invention provides a method for producing a detector non-uniform data set, including: Obtain a detector response image and a scene image; Determine an ideal pixel value according to the detector response image, and preprocess the scene image to obtain an original image; Construct a data degradation model according to the ideal pixel value, perform data degradation processing on the original image according to the data degradation model to obtain a noise image, and construct a detector non-uniform data set according to the noise image and the original image.

[0007] In a possible implementation manner, obtaining the detector response image includes: Perform uniform background shooting under a preset shooting gradient to obtain a plurality of detector response images.

[0008] In a possible implementation manner, determining the ideal pixel value according to the detector response image includes: Calculate the pixel average value of the detector response image to obtain the ideal pixel value corresponding to each pixel point under the pixel value of the detector response image.

[0009] In a possible implementation manner, preprocessing the scene image to obtain the original image includes: Adjust the bit width of the scene image to obtain a standard bit width image; Compress the effective pixel values of the standard bit width image to obtain the original image.

[0010] In a possible implementation manner, constructing the data degradation model according to the ideal pixel value includes: Divide the ideal pixel values into several pixel value intervals; Construct an initial linear correction equation for each pixel value interval; Determine the linear gain and offset value of each initial linear correction equation according to the ideal pixel value to obtain a piecewise linear correction equation for each pixel value interval; Determine the piecewise data degradation equation corresponding to each pixel value interval according to the piecewise linear correction equation, and construct a data degradation model according to the piecewise data degradation equation.

[0011] In a possible implementation manner, performing data degradation processing on the original image according to the data degradation model to obtain a noise image includes: Determine the corresponding piecewise data degradation equation according to the pixel interval where the pixel value of each pixel point in the original image is located; Perform data degradation processing on pixel points according to the segmented data degradation equation to obtain noisy pixel points; After iteratively performing data degradation processing on all pixel points, construct a noise image based on the corresponding noisy pixel points.

[0012] In a possible implementation manner, construct a detector non-uniform data set according to the noise image and the original image, including: Use the noise image as training data and the original image as ground truth data to construct a detector non-uniform data set.

[0013] On the other hand, the present invention also provides a device for making a detector non-uniform data set, including: An image acquisition unit for acquiring a detector response image and a scene image; A preprocessing unit for determining ideal pixel values according to the detector response image and preprocessing the scene image to obtain an original image; A data degradation unit for constructing a data degradation model according to the ideal pixel values, performing data degradation processing on the original image according to the data degradation model to obtain a noise image, and constructing a detector non-uniform data set according to the noise image and the original image.

[0014] On the other hand, the present invention also provides an electronic device, including a processor, a memory, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the above-mentioned method for making a detector non-uniform data set is implemented.

[0015] On the other hand, the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned method for making a detector non-uniform data set is implemented.

[0016] The beneficial effects of the present invention are as follows: In the method for making a detector non-uniform data set provided by the present invention, first, a detector response image and a scene image are acquired; then, ideal pixel values are determined according to the detector response image, and the scene image is preprocessed to obtain an original image; finally, a data degradation model is constructed according to the ideal pixel values, the original image is subjected to data degradation processing according to the data degradation model to obtain a noise image, and a detector non-uniform data set is constructed according to the noise image and the original image. By using the data degradation model constructed from the detector response image to perform data degradation processing on the high-quality scene image, the present invention can efficiently and low-costly generate a high-quality detector non-uniform data set containing various non-uniformity features. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those skilled in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0018] Figure 1 It is a schematic flowchart of an embodiment of the method for making a non-uniform data set of a detector provided by the present invention; Figure 2 It is a schematic flowchart of preprocessing a scene image in an embodiment of the present invention; Figure 3 It is a schematic flowchart of constructing a data degradation model in an embodiment of the present invention; Figure 4 It is a schematic flowchart of data degradation processing in an embodiment of the present invention; Figure 5 It is a schematic structural diagram of an embodiment of a device for making a non-uniform data set of a detector provided by the present invention; Figure 6 It is a schematic structural diagram of an embodiment of an electronic device provided by the present invention. Detailed implementation manners

[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of the present invention.

[0020] In the description of the embodiments of the present invention, unless otherwise specified, "a plurality" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example: A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone.

[0021] The descriptions such as "first" and "second" involved in the embodiments of the present invention are only for descriptive purposes, and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Therefore, the technical features defined with "first" and "second" may explicitly or implicitly include at least one such feature.

[0022] Reference to "embodiment" in this document means that the specific features, structures, or characteristics described in connection with the embodiment may be included in at least one embodiment of the present invention. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein may be combined with other embodiments.

[0023] The present invention provides a method, apparatus, electronic device, and medium for producing a detector non-uniform data set, which will be described separately below.

[0024] Figure 1 FIG. is a schematic flowchart of an embodiment of the method for producing a detector non-uniform data set provided by the present invention, as Figure 1 shown, the method for producing a detector non-uniform data set includes: S101. Obtain a detector response image and a scene image; S102. Determine an ideal pixel value according to the detector response image, and preprocess the scene image to obtain an original image; S103. Construct a data degradation model according to the ideal pixel value, perform data degradation processing on the original image according to the data degradation model to obtain a noise image, and construct a detector non-uniform data set according to the noise image and the original image.

[0025] Compared with the prior art, in the method for producing a detector non-uniform data set provided by the present invention, first, a detector response image and a scene image are obtained; then, an ideal pixel value is determined according to the detector response image, and the scene image is preprocessed to obtain an original image; finally, a data degradation model is constructed according to the ideal pixel value, the original image is subjected to data degradation processing according to the data degradation model to obtain a noise image, and a detector non-uniform data set is constructed according to the noise image and the original image. By using the data degradation model constructed from the detector response image to perform data degradation processing on a high-quality scene image, the present invention can efficiently and low-costly generate a high-quality detector non-uniform data set containing various non-uniformity features.

[0026] It should be noted that the present invention is applicable to the process of producing a data set for non-uniformity correction of deep learning methods for various detectors such as visible, near-infrared, short-wave, or medium-long wave. The following embodiments will be described by taking short-wave infrared as an example.

[0027] In some embodiments of the present invention, obtaining a detector response image includes: Performing uniform background shooting under a preset shooting gradient to obtain a plurality of detector response images.

[0028] In some embodiments of the present invention, determining an ideal pixel value according to the detector response image includes: Calculate the pixel average value of the detector response image to obtain the ideal pixel value corresponding to each pixel at the pixel value of the detector response image.

[0029] Specifically, in order to ensure the accuracy of the reduction noise of the original image, it is necessary to first perform uniform background shooting according to a preset shooting gradient to obtain a number of detector response images. For example, for thermal imaging, the thermal radiation size is adjusted through a preset shooting gradient, and for visible or short-wave, the illumination intensity is adjusted through a preset shooting gradient. In the embodiment, a quantum dot short-wave infrared detector to be corrected is used. In an environment where there is no other light source except the light box in a closed space, first adjust the aperture size of the lens. When the brightness of the light box is 100%, the appropriate aperture is when the detector is fully saturated. Then, according to a 5% illumination intensity gradient, gradually decrease the illumination intensity of the light box until 0%, and collect 21 14-bit detector response images of the uniform background shot by the detector from no illumination to exposure.

[0030] Then, the embodiment generates 21 matrices of 512×640 from the response images, and calculates the average pixel of each matrix to obtain the pixel average value of each image, which is used as the ideal pixel value corresponding to each pixel at the pixel value of the response image.

[0031] In some embodiments of the present invention, Figure 2 is a schematic flowchart of the preprocessing of the scene image in the embodiment of the present invention. As Figure 2 shown, preprocessing the scene image to obtain the original image includes: S201. Adjust the bit width of the scene image to obtain a standard bit width image; S202. Compress the effective pixel values of the standard bit width image to obtain the original image.

[0032] Specifically, in order to ensure the scale and diversity of the data, the scene images mainly collect high-quality data from various scenes such as grasslands, lakesides, buildings, and interiors.

[0033] In the process of preprocessing the scene image data, first, the embodiment performs bit width adjustment, converts a picture with a pixel size of 512×640 into a grayscale image with the same bit width as the detector picture to obtain a standard bit width image. And screen the images, screen out the images with effective detailed information, and eliminate the images with poor effects. Finally, compress the effective pixel values of the images to the ideal effective pixel values corresponding to the detector as the original image in the detector non-uniform data set. Among them, the ideal effective pixel range of the detector used in the embodiment is 8192-16383.

[0034] In some embodiments of the present invention, Figure 3 is a schematic flowchart of constructing a data degradation model in the embodiment of the present invention. As Figure 3As shown, a data degradation model is constructed based on ideal pixel values, including: S301. Divide each ideal pixel value into several pixel value intervals; S302. Construct an initial linear correction equation for each pixel value interval; S303. Determine the linear gain and offset value of each initial linear correction equation according to the ideal pixel value, and obtain the piecewise linear correction equation for each pixel value interval; S304. Determine the piecewise data degradation equation corresponding to each pixel value interval according to the piecewise linear correction equation, and construct a data degradation model according to the piecewise data degradation equation.

[0035] Specifically, in the process of designing the data degradation model, the embodiment divides the response curve of the detector into several pixel value intervals according to the ideal pixel value. For each interval, the embodiment adopts the method of piecewise correction and linear interpolation. Based on the two-point correction method of calibration, the linear gain parameter G and the offset value parameter O are used for the operation of pixel values and ideal values In the linear correction within each interval, the embodiment obtains the actual input values and of each pixel under bright field and dark field by collecting the response images of bright field and dark field under uniform background, and then calculates the average values and of bright field and dark field, which are used to calculate the corrected ideal output value. For each pixel point, the calculation formula of its ideal output value is:

[0036]

[0037]

[0038] wherein, the conversion method between the actual input value and the ideal output value is the linear interpolation method, that is, taking different values of each pixel point as the abscissa and values as the ordinate, and the corresponding value of each value can be obtained by linear interpolation, and the result is consistent with the two-point correction method. The calculation formula is expressed as:

[0039] Then, according to the ideal pixel values of each response image obtained in the previous embodiment, the piecewise linear correction equation of each pixel value interval can be determined. Correspondingly, after obtaining the piecewise linear correction equation of each pixel value interval, the corresponding piecewise data degradation equation takes values as the abscissa and Using the value as the ordinate, the corresponding noisy value can be obtained through ideal values by linear interpolation to obtain the corresponding noisy value , and the calculation formula is expressed as:

[0040] Then, the embodiment degrades the piecewise data degradation equations corresponding to each pixel value interval obtained, and constructs the final data degradation model.

[0041] In some embodiments of the present invention, Figure 4 is a schematic flowchart of the data degradation process of the embodiment of the present invention. As Figure 4 shown, performing data degradation processing on the original image according to the data degradation model to obtain a noisy image includes: S401. Determine the corresponding piecewise data degradation equation according to the pixel interval where the pixel value of each pixel point in the original image is located; S402. Perform data degradation processing on the pixel point according to the piecewise data degradation equation to obtain a noisy pixel point; S403. After iteratively performing data degradation processing on all pixel points, construct a noisy image according to the corresponding noisy pixel points.

[0042] Specifically, in the previous steps, the embodiment has divided the pixel values into several pixel intervals in ascending order with the ideal pixel values corresponding to each response image as the abscissa. During the data degradation process, the embodiment traverses each pixel of the original image to be degraded one by one from left to right and from top to bottom. For each pixel point, first determine which two adjacent intervals on the abscissa the pixel value is in, and then use the data degradation equation corresponding to this interval for data degradation processing, that is, perform a linear interpolation operation using the pixel values in two abscissas and the corresponding two matrices as the ordinate to obtain the pixel value of the degraded noisy pixel point. Finally, after traversing all pixel points, the obtained noisy pixel points are combined into a new image to obtain the corresponding noisy image.

[0043] In some embodiments of the present invention, constructing a detector non-uniform data set according to the noisy image and the original image includes: Using the noisy image as training data and the original image as real data to construct a detector non-uniform data set.

[0044] Specifically, after completing the data degradation processing of all the original images, the embodiment inputs all the original images into a folder as the real data set, and saves the obtained noisy images into another folder with the same file name as the original images as the corresponding training data set. The real data set and the training data set are combined into a complete detector non-uniform data set.

[0045] In summary, in order to provide a high-quality dataset for the image non-uniformity correction method based on deep learning efficiently and at low cost, the present invention first obtains a detector response image and a scene image; then determines the ideal pixel value according to the detector response image, and preprocesses the scene image to obtain an original image; finally constructs a data degradation model according to the ideal pixel value, performs data degradation processing on the original image according to the data degradation model to obtain a noise image, and constructs a detector non-uniform dataset according to the noise image and the original image. By using the data degradation model constructed from the detector response image to perform data degradation processing on the high-quality scene image, the present invention can efficiently and at low cost generate a high-quality detector non-uniform dataset containing various non-uniformity features.

[0046] To better implement the method for manufacturing a detector non-uniform dataset in the embodiments of the present invention, correspondingly, based on the method for manufacturing a detector non-uniform dataset, as Figure 5 shown, the present invention also provides a device 500 for manufacturing a detector non-uniform dataset. The device 500 for manufacturing a detector non-uniform dataset includes: An image acquisition unit 501, configured to acquire a detector response image and a scene image; A preprocessing unit 502, configured to determine an ideal pixel value according to the detector response image, and preprocess the scene image to obtain an original image; A data degradation unit 503, configured to construct a data degradation model according to the ideal pixel value, perform data degradation processing on the original image according to the data degradation model to obtain a noise image, and construct a detector non-uniform dataset according to the noise image and the original image.

[0047] The device 500 for manufacturing a detector non-uniform dataset provided in the above embodiments can implement the technical solutions described in the embodiments of the method for manufacturing a detector non-uniform dataset. For the specific implementation principles of the above modules or units, reference can be made to the corresponding content in the embodiments of the method for manufacturing a detector non-uniform dataset, which will not be elaborated here.

[0048] As Figure 6 shown, the present invention also correspondingly provides an electronic device 600. The electronic device 600 includes a processor 601, a memory 602, and a display 603. Figure 6 Only some components of the electronic device 600 are shown, but it should be understood that it is not required to implement all the shown components, and more or fewer components can be alternatively implemented.

[0049] The processor 601 may be a central processing unit (CPU), a microprocessor, or other data processing chips in some embodiments, and is configured to run the program code stored in the memory 602 or process data, such as the method for manufacturing a detector non-uniform dataset in the present invention.

[0050] In some embodiments, the processor 601 may be a single server or a server group. The server group may be centralized or distributed. In some embodiments, the processor 601 may be local or remote. In some embodiments, the processor 601 may be implemented on a cloud platform. In one embodiment, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-cloud, etc., or any combination thereof.

[0051] The memory 602 may be an internal storage unit of the electronic device 600 in some embodiments, such as a hard disk or memory of the electronic device 600. The memory 602 may also be an external storage device of the electronic device 600 in other embodiments, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. equipped on the electronic device 600.

[0052] Furthermore, the memory 602 may include both an internal storage unit and an external storage device of the electronic device 600. The memory 602 is used to store application software installed on the electronic device 600 and various types of data.

[0053] The display 603 may be an LED display, a liquid crystal display, a touch liquid crystal display, and an OLED (Organic Light-Emitting Diode) toucher, etc. in some embodiments. The display 603 is used to display information of the electronic device 600 and to display a visual user interface. The components 601 - 603 of the electronic device 600 communicate with each other through a system bus.

[0054] In one embodiment, when the processor 601 executes the detector non-uniform data set making program in the memory 602, the following steps may be implemented: Obtain a detector response image and a scene image; Determine an ideal pixel value according to the detector response image, and preprocess the scene image to obtain a raw image; Construct a data degradation model according to the ideal pixel value, perform data degradation processing on the raw image according to the data degradation model to obtain a noise image, and construct a detector non-uniform data set according to the noise image and the raw image.

[0055] It should be understood that when the processor 601 executes the detector non-uniform data set making program in the memory 602, in addition to the above functions, other functions may also be implemented. For specific details, reference may be made to the description of the corresponding method embodiments above.

[0056] Accordingly, an embodiment of the present application further provides a computer-readable storage medium, which is used to store computer-readable programs or instructions. When the programs or instructions are executed by a processor, the steps or functions in the method for producing a detector non-uniform data set provided by the above various method embodiments can be implemented.

[0057] Those skilled in the art can understand that all or part of the processes for implementing the methods of the above embodiments can be completed by instructing relevant hardware (such as a processor, a controller, etc.) through a computer program, and the computer program can be stored in a computer-readable storage medium. Among them, the computer-readable storage medium is a magnetic disk, an optical disk, a read-only memory or a random access memory, etc.

[0058] The method, device, electronic device and storage medium for producing a detector non-uniform data set provided by the present invention have been introduced in detail above. Specific examples are used in this article to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those skilled in the art, according to the idea of the present invention, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for producing a non-uniform data set of a detector, characterized in that, including: Obtain a detector response image and a scene image; Determine an ideal pixel value according to the detector response image, and preprocess the scene image to obtain an original image; Construct a data degradation model according to the ideal pixel value, perform data degradation processing on the original image according to the data degradation model to obtain a noise image, and construct a detector non-uniform data set according to the noise image and the original image.

2. The method for producing a non-uniform data set of a detector according to claim 1, wherein The obtaining of the detector response image includes: Perform uniform background shooting under a preset shooting gradient to obtain a plurality of detector response images.

3. The method for producing a non-uniform data set of detectors according to claim 1, characterized in that, The determining of the ideal pixel value according to the detector response image includes: Calculate the pixel average value of the detector response image to obtain the ideal pixel value corresponding to each pixel point under the pixel value of the detector response image.

4. The method for producing a non-uniform data set of a detector according to claim 1, wherein The preprocessing of the scene image to obtain the original image includes: Adjust the bit width of the scene image to obtain a standard bit width image; Compress the effective pixel values of the standard bit width image to obtain an original image.

5. The method for producing a non-uniform data set of detectors according to claim 1, characterized in that, The constructing of the data degradation model according to the ideal pixel value includes: Divide the ideal pixel values into a plurality of pixel value intervals; Construct an initial linear correction equation for each pixel value interval; Determine the linear gain and offset value of each initial linear correction equation according to the ideal pixel value to obtain a piecewise linear correction equation for each pixel value interval; Determine a piecewise data degradation equation corresponding to each pixel value interval according to the piecewise linear correction equation, and construct a data degradation model according to the piecewise data degradation equation.

6. The method for producing a non-uniform data set of detectors according to claim 5, characterized in that, The performing of data degradation processing on the original image according to the data degradation model to obtain a noise image includes: Determine a corresponding piecewise data degradation equation according to the pixel interval where the pixel value of each pixel point in the original image is located; Perform data degradation processing on the pixel point according to the piecewise data degradation equation to obtain a noisy pixel point; After iteratively performing data degradation processing on all pixel points, construct a noise image according to the corresponding noisy pixel points.

7. The method for producing a non-uniform data set of detectors according to claim 1, characterized in that, The constructing of the detector non-uniform data set according to the noise image and the original image includes: Use the noise image as training data and the original image as real data to construct a detector non-uniform data set.

8. A device for making a non-uniform data set of a detector, characterized in that, including: An image acquisition unit for obtaining a detector response image and a scene image; A preprocessing unit for determining an ideal pixel value according to the detector response image and preprocessing the scene image to obtain an original image; A data degradation unit for constructing a data degradation model according to the ideal pixel value, performing data degradation processing on the original image according to the data degradation model to obtain a noise image, and constructing a detector non-uniform data set according to the noise image and the original image.

9. An electronic device, comprising a processor, a memory, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method for manufacturing a detector non-uniform data set according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for manufacturing a detector non-uniform data set according to any one of claims 1 to 7.

Images