Detection method and system for complex ground random target

Abstract

The invention discloses a detection method and system for a complex ground random target, and the method comprises the steps: carrying out the data collection of a complex ground through collection equipment, and obtaining initial collection data; the data acquisition device at least comprises a hyperspectral acquisition device, an SAR acquisition device and a camera acquisition device; performing feature marking on the initial acquisition data acquired by the camera acquisition equipment; if the feature mark exists, carrying out detection and identification to obtain a detection result, and stopping carrying out detection and identification on initial acquisition data acquired by the hyperspectral acquisition equipment and the SAR acquisition equipment; and if no feature mark exists, performing consistency processing, normalization processing and detection identification on the initial acquisition data acquired by the hyperspectral acquisition device and the SAR acquisition device to obtain a detection result. In the invention, two channels are identified and detected, the identification efficiency is not influenced, and the identification cost is obviously reduced.

Description

Technical Field

[0001] The invention belongs to the field of target detection, and in particular relates to a detection method and system for complex ground random targets. Background Art

[0002] With the rapid development of target detection technology, many detection and identification technologies have emerged, such as aperture radar (SAR), optical, infrared, hyperspectral and lidar detection methods, and in order to ensure detection accuracy, detection methods such as model algorithms have emerged. However, due to the large amount of data in the model algorithm, its processing process depends on the cloud / server equipment for analysis and calculation, which takes a long time, and conventional terminals cannot meet the computing power requirements. However, during the detection process, the detection target is random and difficult to determine, and low-feature or high-feature target objects may appear. If complex model algorithms are used, there will be waste and the detection time will be long. Summary of the invention

[0003] In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is: how to realize target detection at low cost, quickly and efficiently.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0005] On the one hand, a detection method for complex ground random targets, the key technology of which is: the detection method is:

[0006] The acquisition equipment collects data from the complex ground to obtain initial acquisition data; wherein the acquisition equipment for data acquisition at least includes: a hyperspectral acquisition equipment, a SAR acquisition equipment, and a camera acquisition equipment;

[0007] Performing feature marking on the initial acquisition data acquired by the camera acquisition device;

[0008] If there are characteristic marks, detection and identification are performed to obtain detection results, and detection and identification are performed on the initial data collected by the hyperspectral acquisition equipment and the SAR acquisition equipment;

[0009] If there is no characteristic mark, the initial acquisition data acquired by the hyperspectral acquisition device and the SAR acquisition device are subjected to consistency processing, normalization processing, detection and identification to obtain a detection result.

[0010] A further technical solution is that the initial acquisition data of the hyperspectral acquisition device, the SAR acquisition device, and the camera acquisition device all have acquisition images;

[0011] All the acquisition devices have the same acquisition cycle, and all images are correspondingly provided with a cycle label, and each acquisition device acquires at least one image in each acquisition cycle;

[0012] Images with the same period label overlap.

[0013] A further technical solution is: the pictures collected by the camera acquisition device are input into the terminal detection module for feature marking, screening processing, detection and identification;

[0014] The data collected by the hyperspectral acquisition device and the SAR acquisition device are input into the cloud for consistency processing, normalization processing, detection and identification; and the identification results are fed back to the terminal.

[0015] A further technical solution is: a specific method for marking the initial data collected by the camera acquisition device:

[0016] Compare and identify the camera images collected by the camera acquisition device with the existing features;

[0017] If the overlap degree is higher than the preset threshold, the overlap area is adaptively selected and the detection result is output.

[0018] A further technical solution is: the method for detecting and identifying the initial data collected by the hyperspectral collection device and the SAR collection device is as follows;

[0019] Locking the data collected by the hyperspectral acquisition device and the SAR acquisition device and the sequence label according to the periodic sequence label of the camera image with the characteristic mark;

[0020] The terminal terminates the transmission channel corresponding to the hyperspectral acquisition device and the SAR acquisition device for collecting data to the cloud, or terminates the acquisition program of the hyperspectral acquisition device and the SAR acquisition device.

[0021] A further technical solution is: the data collected by the hyperspectral acquisition device and the SAR acquisition device are input into the cloud for consistency processing, normalization processing, and detection and identification.

[0022] Based on the pixel aggregation resampling method conforming to the linear mixed model, the data collected by the SAR acquisition equipment is downsampled to obtain SAR downsampled data;

[0023] Perform data normalization on the data collected by the hyperspectral acquisition equipment and the SAR downsampled data

[0024] Perform pixel-level registration processing, establish a pixel-by-pixel logical relationship between the output results of the hyperspectral acquisition equipment and SAR acquisition equipment, perform decision-fusion detection analysis, and obtain detection results.

[0025] A further technical solution is: the pixel-level registration processing is performed, a pixel-by-pixel logical relationship between the output results of the hyperspectral acquisition device and the SAR acquisition device is established, and a decision fusion detection analysis is performed to obtain the detection result, and the method is:

[0026] According to the feature point selection corresponding to the reference image and the image to be registered, the transformation parameters between the two images are solved;

[0027] Using the affine transformation algorithm, the image to be registered is transformed into a corresponding space so that the two images are in the same coordinate space coordinate system;

[0028] The pixel aggregation algorithm is used to reassign the pixel values ​​of the image to be registered after spatial transformation through grayscale transformation, that is, resampling.

[0029] Adopt the idea of ​​winner-takes-all and carry out integration.

[0030] On the other hand, a detection system for complex ground random targets is proposed, comprising a terminal; the terminal is connected to a hyperspectral acquisition device, a SAR acquisition device, and a camera acquisition device;

[0031] A terminal detection module is provided in the terminal, and the terminal detection module is used to obtain the camera image collected by the camera acquisition device;

[0032] The terminal is provided with a cloud communication module, and the terminal is connected to the cloud via the cloud communication module; the images collected by the hyperspectral acquisition device and the SAR acquisition device are sent to the cloud for detection and identification, and the terminal obtains the cloud detection and identification results via the cloud communication module.

[0033] Further technical content is: a transmission cut-off module is provided in the terminal to cut off the communication channel with the cloud.

[0034] Further technical content is: a transmission cut-off module is arranged in the terminal, which is used to cut off the acquisition program acquired by the hyperspectral acquisition device and the SAR acquisition device.

[0035] The beneficial effects of the present invention include:

[0036] The terminal and the cloud are combined to detect and identify targets. The terminal can identify targets with obvious features. For targets with low features and unrecognizable targets, the cloud uses model algorithms for accurate identification. After calculation and comparison, the recognition efficiency is not affected and the recognition cost is significantly reduced.

[0037] And by utilizing the time difference between the acquisition time of the camera acquisition device and the hyperspectral acquisition device and the SAR acquisition device, the acquisition time of the camera acquisition device is fast, while the acquisition time of the hyperspectral acquisition device and the SAR acquisition device is long. After the acquisition and recognition of the picture collected by the camera acquisition device is completed, if the recognition is successful, the acquisition program of the hyperspectral acquisition device and the SAR acquisition device can be terminated, or the data transmission channel sent to the cloud can be terminated, or a stop recognition program can be sent to the cloud;

[0038] When cloud recognition and terminal recognition are carried out in parallel, terminal recognition has less impact on the hyperspectral acquisition equipment, SAR acquisition equipment, data transmission to the cloud, and target recognition procedures in the cloud. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0040] Figure 1 This is an overall diagram of the detection and identification process of the present invention.

[0041] Figure 2 This is a schematic diagram of the image data fusion processing effect of the present invention;

[0042] Figure 3 is a block diagram of the detection system of the present invention;

[0043] Figure 4 It is a schematic diagram of the hardware structure of the electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0045] In the prior art, if synthetic aperture radar (SAR), optical, infrared, hyperspectral and lidar are used for target detection and identification, the acquisition cycle plus the recognition process cycle will result in a longer cycle.

[0046] Therefore, see Figure 1 The embodiment of the present invention provides a detection method for complex ground random targets, the detection method is:

[0047] S1: The hyperspectral acquisition device, the SAR acquisition device, and the video acquisition device simultaneously acquire data on the complex ground, wherein the video acquisition device acquires acquisition pictures, the hyperspectral acquisition device acquires hyperspectral acquisition data, and the SAR acquisition device acquires SAR acquisition data; wherein the hyperspectral acquisition data and the SAR acquisition data both contain picture data.

[0048] S2: feature marking the initial data collected by the camera acquisition device;

[0049] S3: Determine whether the initial acquisition data acquired by the camera acquisition device has a characteristic mark, if so, proceed to step S4; otherwise, proceed to step S5;

[0050] S4: Perform detection and identification to obtain detection results, and stop detecting and identifying the initial data collected by the hyperspectral acquisition device and the SAR acquisition device;

[0051] S5: performing consistency processing, normalization processing, detection and identification on the initial data collected by the hyperspectral collection device and the SAR collection device to obtain a detection result.

[0052] In this embodiment, two channels are proposed for target recognition and detection, one is a channel-type terminal recognition channel, and the other is a terminal acquisition cloud recognition channel;

[0053] In step S1, the acquisition cycles of the hyperspectral acquisition device, the SAR acquisition device, and the camera acquisition device are the same, and all acquisition category images are correspondingly provided with cycle labels, and the hyperspectral acquisition device, the SAR acquisition device, and the camera acquisition device all acquire at least one image in each acquisition cycle;

[0054] In this embodiment, before acquisition, the hyperspectral acquisition device, SAR acquisition device, and camera acquisition device need to be pre-debugged to ensure that the image data acquired by the three have overlapping parts, and the overlapping areas are marked. The images acquired by each type of acquisition device and in the same cycle have overlaps. In this embodiment, the overlapping part is not less than 70%.

[0055] In this embodiment, the pictures collected by the camera acquisition device are input into the terminal detection module for feature marking, screening processing, detection and identification;

[0056] The data collected by the hyperspectral acquisition device and the SAR acquisition device are input into the cloud for consistency processing, normalization processing, detection and identification; and the identification results are fed back to the terminal.

[0057] In S2, the specific method of performing feature marking on the initial acquisition data acquired by the camera acquisition device is as follows:

[0058] S21: Compare and identify the camera image collected by the camera collection device with the existing features;

[0059] S22: If the overlap degree is higher than a preset threshold, the overlap area is adaptively selected and the detection result is output.

[0060] In this embodiment, the terminal is provided with a target feature database, and in step S21, the camera image and the existing feature data in the target feature database are stored. In order to ensure the recognition speed and reduce the memory usage of the terminal, the target feature database stores no more than 500 features, and the comparison is performed in parallel + serial. If there is a target classification, the categories can be selected in parallel according to the categories and then compared one by one within the categories. For example, 500 copies are divided into 10 categories, each with 10-100 copies; if the preliminary comparison belongs to the 8th category of features, they are compared one by one with the feature pictures in the 8th category of features.

[0061] In step S4, the method for detecting and identifying the initial data collected by the hyperspectral collection device and the SAR collection device is as follows:

[0062] Locking the data collected by the hyperspectral acquisition device and the SAR acquisition device and the sequence label according to the periodic sequence label of the camera image with the characteristic mark;

[0063] The terminal terminates the acquisition program of the hyperspectral acquisition device and the SAR acquisition device;

[0064] Or the terminal cuts off the transmission channel corresponding to the data collected by the hyperspectral acquisition device and the SAR acquisition device to the cloud;

[0065] Or the terminal sends a cut-off identification program signal to the cloud.

[0066] Based on the above three situations, it depends on the terminal identification channel completing the identification detection program, and the terminal collects the current node of the cloud identification channel.

[0067] 1. If the terminal collection cloud recognition channel is in the stage of data collection by the hyperspectral collection device and the SAR collection device, if the target has been identified through the terminal recognition channel, the hyperspectral collection device and the SAR collection device can be stopped from collecting data;

[0068] 2. If the terminal collection cloud identification channel is at a node where the terminal transmits data to the cloud, the terminal cuts off the transmission channel corresponding to the hyperspectral collection device and the SAR collection device collecting data to the cloud;

[0069] 3. If the cloud recognition channel collected by the terminal is in the cloud recognition state, the terminal sends a termination recognition program signal to the cloud, and the cloud stops recognition detection.

[0070] In S5, combined Figure 2 It can be seen that the method of inputting the data collected by the hyperspectral acquisition device and the SAR acquisition device into the cloud for consistency processing, normalization processing, and detection and identification is:

[0071] S51: Based on the pixel aggregation resampling method in accordance with the linear mixed model, the data collected by the SAR acquisition device is downsampled to obtain SAR downsampled data; its resolution is consistent with the grayscale resolution of the output of the hyperspectral data detection; specifically, the pixel aggregation method is used to downsample the data collected by the SAR acquisition device, and the surrounding pixel values ​​that contribute to the output pixel are weighted averaged according to the area to obtain the value of the new pixel. See the attached Figure 2 ; For pixel b1, the contributing pixels are a1, a2, a4, and a5. Therefore, when the image is reduced from 3×3 to 2×2, the spatial resolution is 1.5 times the original one. The calculation method is:

[0072] b1=(a1×100%+a2×50%+a4×50%+a5×25%) / (1.5×1.5)

[0073] S52: performing data normalization processing on the data collected by the hyperspectral collection device and the SAR down-sampling data, so that the grayscale values ​​of the two grayscale images finally obtained are within the same magnitude range;

[0074]

[0075] in,

[0076] output_x i The data collected by the hyperspectral acquisition device after normalization or the SAR down-sampling data;

[0077] input_x i For: data collected by hyperspectral acquisition equipment or SAR down-sampled data;

[0078] min(x1,...,x n-1 ,x n ,) is: the minimum value of the data collected by the hyperspectral acquisition device or the SAR down-sampling data;

[0079] max(x1,...,x n-1 ,x n ,) is: the maximum value of the data collected by the hyperspectral acquisition device or the SAR down-sampling data.

[0080] S53: Perform pixel-level registration processing to obtain pixel-by-pixel output results of the hyperspectral acquisition equipment and SAR acquisition equipment matching the geographic coordinates, adopt the winner-takes-all concept, perform decision-making fusion detection analysis, and obtain detection results. Specifically, the specific content of S53 is:

[0081] S531: according to the feature point selection corresponding to the reference image and the image to be registered, such as the intersection of the road, the obvious mountain top, the fork of the river, etc., solve the transformation parameters between the two images;

[0082] S532: Using an affine transformation algorithm, the images to be registered are subjected to corresponding spatial transformation, so that the two images are in the same coordinate space coordinate system.

[0083] The affine transformation algorithm is detailed on the web page:

[0084] https: / / yebd1h.smartapps.cn / pages / blog / index?blogId=136283646&_ refluxos=a10&_swebfr=26&_swebFromHost=bdhonorbrowser .

[0085] S533: using a pixel aggregation algorithm to reassign pixel values ​​of the image to be registered after spatial transformation through grayscale transformation, i.e., resampling;

[0086] For details on the pixel aggregation algorithm, please refer to the doctoral dissertation: Hyperspectral detection model and spatial-spectral sensitivity analysis of targets in shadow areas - Gao Yingqian - Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences.

[0087] The winner-takes-all concept is adopted to fuse the hyperspectral and SAR detection and recognition results. The calculation method is:

[0088]

[0089] input_x n，n Indicates: the nth row and nth column pixel of the detection result (image matrix);

[0090] input_x n Indicates: the nth column pixel of the image matrix;

[0091] input_xi n,n Represents: the pixel in the nth row and nth column of the i-th image.

[0092] Combination Figure 2 It can be seen that in the figure, the i small pictures in the first row are from left to right:

[0093] 1. The pixel input_x1 in the first row and first column of the first image 1,1 ;

[0094] 2. The pixel input_x2 in the first row and first column of the second image 1,1 ;

[0095] …

[0096] i, the pixel input_xi in the first row and first column of the i-th image 1,1 ;

[0097] The second row is the pixels in the first row and first column of the detection result (image matrix).

[0098] For the single use of SAR target detection technology, SAR target detection has the characteristics of high resolution, all-day and all-weather monitoring and strong penetration. The reflection characteristics of ground objects in the high spatial resolution SAR images obtained are affected by microwave frequency, reflection angle, polarization mode and terrain, resulting in different representations of the same object, which is not conducive to image interpretation;

[0099] For the single use of hyperspectral remote sensing target detection technology, hyperspectral remote sensing target detection has rich information, accurate target identification, and strong environmental monitoring capabilities, but hyperspectral satellite sensors are completely dependent on light sources, cannot image at night, and are easily affected by bad weather, resulting in the loss of some obvious target information in the hyperspectral image, and the data processing volume is large and the technical requirements are high;

[0100] Traditional SAR and hyperspectral image fusion algorithms can be divided into three categories:

[0101] 1. Component Substitution (CS) method, mainly including principal component analysis (PCA), IHS transformation and Brovey, etc.; this type of algorithm is simple to implement, and the SAR image detail information is obvious in the fusion result. However, due to the large difference in the spectral range between SAR images and hyperspectral images, simply replacing the SAR image with the principal component of the hyperspectral image will cause serious spectral distortion in the fusion result.

[0102] 2. Multi-resolution analysis (MRA) methods, including wavelet method, pyramid transform, non-subsampled contourlet transform (NSCT), non-subsampled shearlet transform (NSST), etc. For example, Song Jianshe et al. used the wavelet transform method to fuse SAR and hyperspectral images. The spectrum of the fused image was improved, but there were artifacts and edge jagged edges; Su Zhiyuan et al. made full use of the translation invariance, multi-scale, multi-directionality and strong contour information expression ability of NSCT, and proposed a SAR and hyperspectral image fusion algorithm based on NSCT, which significantly improved the fusion effect. Compared with the CS method, the MRA method can effectively suppress the noise of SAR images and reduce spectral distortion, but it is easily affected by image registration, resulting in aliasing and edge artifacts.

[0103] 3. Hybrid method, that is, using CS method to separate the intensity information and spectral information of hyperspectral image, applying MRA method to fuse the intensity components of SAR and hyperspectral image to obtain new components, and finally inverse transforming the new components and spectral information to obtain fused image. For example, Mercer et al. use SAR and hyperspectral image fusion algorithm based on hybrid IHS transform and wavelet transform, and replace the approximate components of SAR image after wavelet decomposition with the approximate components in the intensity components of hyperspectral image to enhance the spatial detail information of multispectral. This algorithm can obtain better spatial and spectral information, but the CS method is limited to fusion methods that can perform spatial transformation (such as PCA and IHS).

[0104] Based on this, the present invention proposes a target detection method that integrates hyperspectral and SAR decision fusion, and utilizes the decision fusion of hyperspectral and SAR detection and recognition results to realize deep mining and detection and recognition of low-feature targets under complex ground background conditions.

[0105] The method proposed in the present invention efficiently constructs a decision fusion model for hyperspectral and SAR detection and recognition results, performs downsampling processing on SAR detection result data and data normalization processing on detection and recognition results of hyperspectral and SAR data, establishes a new method and system for detecting and recognizing low-feature targets such as low scattering, weak radiation, and micro-reflection in a scene where the target is fused with a complex ground background environment, scientifically guides the deep recognition of target features under complex ground background conditions, efficiently improves the detection and recognition of low-feature non-cooperative targets and the design and effectiveness evaluation efficiency of low-feature cooperative targets under the new situation, and significantly reduces the detection, recognition and precision guidance costs of reconnaissance / strike platforms such as helicopters and UAVs under air-to-ground remote sensing imaging conditions.

[0106] On the other hand, combined Figure 3 It can be shown that a detection system for complex ground random targets is provided, including a terminal; the terminal can be mounted on a drone or a collection aircraft or other collection and mounting equipment to achieve close-range target recognition and detection, and serve as a cloud recognition transmission mid-terminal to perform data transmission and obtain recognition results. For example, a terminal on a helicopter, etc.

[0107] It can also be a ground operation control terminal, which is used to implement ground monitoring detection and recognition to obtain remote recognition results, such as the relationship between the drone and the control terminal;

[0108] The terminal is connected to a hyperspectral acquisition device, a SAR acquisition device, and a camera acquisition device;

[0109] The connection method may be a wired or wireless connection, which is not limited here.

[0110] A terminal detection module is provided in the terminal, and the terminal detection module is used to obtain the camera image collected by the camera acquisition device;

[0111] The terminal is provided with a cloud communication module, and the terminal is connected to the cloud via the cloud communication module; the images collected by the hyperspectral acquisition device and the SAR acquisition device are sent to the cloud for detection and identification, and the terminal obtains the cloud detection and identification results via the cloud communication module.

[0112] In this embodiment, a cutoff module is provided in the terminal for cutting off the communication channel with the cloud.

[0113] In this embodiment, a cutoff module is provided in the terminal for cutting off the acquisition program acquired by the hyperspectral acquisition device and the SAR acquisition device.

[0114] In this embodiment, the video acquisition device adopts a pan-tilt camera.

[0115] Figure 4 is a schematic diagram of the structure of an electronic device provided by an optional embodiment of the present invention, such as Figure 4 As shown, the electronic device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. For example, it is used to connect a hyperspectral acquisition device, a SAR acquisition device, and a camera acquisition device. The various components are connected to each other using different buses for communication, and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed in the electronic device, including instructions stored in or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple electronic devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 4 A processor 10 is taken as an example.

[0116] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0117] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0118] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0119] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0120] The electronic device further includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 4 The example of connecting through bus is taken in the following.

[0121] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the electronic device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator rod, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0122] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0123] The above are only preferred implementations of the present invention. It should be pointed out that a number of modifications and improved technical solutions made by those skilled in the art without departing from the technical solution should also be deemed to fall within the scope of protection required by the claims.

Claims

1. A detection method for complex ground random targets, characterized in that: The detection method is: The acquisition equipment collects data on the complex ground to obtain initial acquisition data; The data acquisition equipment includes at least: a hyperspectral acquisition device, a SAR acquisition device, and a camera acquisition device; Performing feature marking on the initial acquisition data acquired by the camera acquisition device; If there are characteristic marks, detection and identification are performed to obtain detection results, and detection and identification are performed on the initial data collected by the hyperspectral acquisition equipment and the SAR acquisition equipment; If there is no characteristic mark, the initial acquisition data acquired by the hyperspectral acquisition device and the SAR acquisition device are subjected to consistency processing, normalization processing, detection and identification to obtain a detection result.

2. The method according to claim 1, characterized in that: The initial acquisition data of the hyperspectral acquisition device, SAR acquisition device, and camera acquisition device all include acquisition images; All the acquisition devices have the same acquisition cycle, and all images are correspondingly provided with a cycle label, and each acquisition device acquires at least one image in each acquisition cycle; Images with the same period label overlap.

3. The method according to claim 2, characterized in that: The images collected by the camera acquisition device are input into the terminal detection module for feature marking, screening, and detection and identification; The data collected by the hyperspectral acquisition device and the SAR acquisition device are input into the cloud for consistency processing, normalization processing, and detection and identification; And feed back the recognition results to the terminal.

4. The method according to claim 3, characterized in that: The specific method for feature marking the initial data collected by the camera acquisition device is as follows: Compare and identify the camera images collected by the camera acquisition device with the existing features; If the overlap degree is higher than the preset threshold, the overlap area is adaptively selected and the detection result is output.

5. The method according to claim 4, characterized in that: The method for detecting and identifying the initial data collected by the hyperspectral collection device and the SAR collection device is as follows: Locking the data collected by the hyperspectral acquisition device and the SAR acquisition device and the sequence label according to the periodic sequence label of the camera image with the characteristic mark; The terminal terminates the transmission channel corresponding to the hyperspectral acquisition device and the SAR acquisition device for collecting data to the cloud, or terminates the acquisition program of the hyperspectral acquisition device and the SAR acquisition device.

6. The method according to claim 3, characterized in that: The method of inputting the data collected by the hyperspectral acquisition device and the SAR acquisition device into the cloud for consistency processing, normalization processing, and detection and identification is as follows: Based on the pixel aggregation resampling method conforming to the linear mixed model, the data collected by the SAR acquisition equipment is downsampled to obtain SAR downsampled data; Perform data normalization on the data collected by the hyperspectral acquisition equipment and the SAR downsampled data Perform pixel-level registration processing, establish a pixel-by-pixel logical relationship between the output results of the hyperspectral acquisition equipment and SAR acquisition equipment, perform decision-fusion detection analysis, and obtain detection results.

7. The method according to claim 6, characterized in that: The pixel-level registration processing is performed to establish a pixel-by-pixel logical relationship between the output results of the hyperspectral acquisition device and the SAR acquisition device, and a decision fusion detection analysis is performed to obtain the detection result. The method is: According to the feature point selection corresponding to the reference image and the image to be registered, the transformation parameters between the two images are solved; Using the affine transformation algorithm, the image to be registered is transformed into a corresponding space so that the two images are in the same coordinate space coordinate system; The pixel aggregation algorithm is used to reassign the pixel values ​​of the image to be registered after spatial transformation through grayscale transformation, that is, resampling. Adopt the idea of ​​winner-takes-all and carry out integration.

8. A detection system for complex ground random targets, characterized by: It includes a terminal; the terminal is connected to a hyperspectral acquisition device, a SAR acquisition device, and a camera acquisition device; A terminal detection module is provided in the terminal, and the terminal detection module is used to obtain the camera image collected by the camera acquisition device; The terminal is provided with a cloud communication module, and the terminal is connected to the cloud via the cloud communication module; the images collected by the hyperspectral acquisition device and the SAR acquisition device are sent to the cloud for detection and identification, and the terminal obtains the cloud detection and identification results via the cloud communication module.

9. The system according to claim 8, characterized in that: The terminal is provided with a cut-off module for cutting off the communication channel with the cloud.

10. The system according to claim 8, characterized in that: The terminal is provided with a cut-off module for cutting off the acquisition program acquired by the hyperspectral acquisition device and the SAR acquisition device.

Images