A target detection method, device, storage medium and electronic device

Abstract

This specification discloses a target detection method, device, storage medium, and electronic device. The target detection method provided in this specification can obtain the reflected light of a distant target as incident light, process the incident light using a telescopic imaging group, perform diffused imaging on the processed incident light through a defocus lens group, then obtain a spectral compression image through a area array detector. Finally, use the pre-trained target detection model in the target detection unit to perform target detection on the spectral compression image. This method can diffusely magnify the spectral compression image of a distant target through the defocus lens group, facilitating target positioning of the distant target, while also making the characteristic information of the distant target more abundant, facilitating the identification of the type of the distant target, and improving the accuracy of the detection result of the distant target.

Description

Technical Field

[0001] This specification relates to the field of image recognition, and particularly to an object detection method, apparatus, storage medium, and electronic device. Background Art

[0002] With the progress of technology, existing image detection technologies have become relatively mature. Through image detection technologies, many objects can be recognized and located.

[0003] Currently, object detection of distant objects is a difficult problem in the field of image recognition. For distant objects, the object images in the captured images are often small, resulting in the object detection system being unable to effectively recognize distant objects. Although the spectral compression reconstruction technology can effectively obtain hyperspectral images, there is a certain error between the reconstructed image and the true object image, resulting in a reduction in the accuracy of the object detection results. Summary of the Invention

[0004] This specification provides an object detection method, apparatus, storage medium, and electronic device to partially solve the above problems existing in the prior art.

[0005] This specification adopts the following technical solutions:

[0006] This specification provides an object detection method. The object detection method performs object detection on a distant object through a spectral compression imaging system. The spectral compression imaging system includes: a telescopic imaging group, a defocus lens group, a planar array detector, and an object detection unit. The telescopic imaging group includes: an objective lens and an eyepiece, including:

[0007] Obtain the reflected light of the distant object as the incident light;

[0008] Perform aperture compression on the incident light through the objective lens and the eyepiece in the telescopic imaging group;

[0009] Process the incident light after aperture compression through the defocus lens group to expand the imaging area of the object, and obtain a spectral compression image with the expanded imaging area of the object through the planar array detector;

[0010] Input the spectral compression image into a pre-trained object detection model deployed in the object detection unit to perform object detection on the spectral compression image through the object detection model.

[0011] Optionally, the spectral compression imaging system further includes: a diffractive optical element;

[0012] Before processing the incident light after aperture compression through the defocus lens group, the method further includes:

[0013] Amplitude and phase modulation are performed on different spectra corresponding to the incident light at different wavelengths through the diffractive optical element to obtain multiple spectra of different wavelengths;

[0014] The incident light after aperture compression is processed through the defocus lens group to expand the imaging area of the target object. Specifically, it includes:

[0015] The defocus lens group processes multiple spectra of different wavelengths obtained after modulation by the diffractive optical element to expand the imaging areas of the spectra of the target object at different wavelengths;

[0016] The spectral compression image of the target object after the imaging area is expanded is obtained through the area array detector. Specifically, it includes:

[0017] The area array detector integrates and accumulates the expanded imaging areas of the spectra of the target object at different wavelengths to obtain the spectral compression image of the target object.

[0018] Optionally, the incident light after aperture compression is processed through the defocus lens group to expand the imaging area of the target object. Specifically, it includes:

[0019] The defocus lens group performs defocus processing on the incident light after aperture compression to distinguish the imaging area of the target object from the background and expand the imaging area of the target object.

[0020] Optionally, the spectral compression image is input into a pre-trained target detection model deployed in the target detection unit to perform target detection on the spectral compression image through the target detection model. Specifically, it includes:

[0021] The spectral compression image is input into a pre-trained target detection model deployed in the target detection unit to output a heat map for the spectral compression image through the target detection model, and the heat map is used to represent the coordinate information of the target object and the target information of the target object;

[0022] Based on the heat map, a target detection result is obtained.

[0023] Optionally, training the target detection model specifically includes:

[0024] Obtain sample images;

[0025] Input the sample image into the target detection model to be trained, and perform target detection on the sample image through the target detection model to obtain a heat map to be optimized;

[0026] Taking the minimization of the deviation between the coordinate information corresponding to the image of the sample target in the sample image included in the heat map to be optimized and the label coordinates corresponding to the sample image, and the deviation between the target information corresponding to the image of the sample target in the sample image included in the heat map to be optimized and the label target information corresponding to the sample image as the optimization objective, train the target detection model.

[0027] Optionally, obtaining a sample image specifically includes:

[0028] Obtain each sample image for the same sample target, where different sample images correspond to spectra of different wavelengths;

[0029] Inputting the sample image into the target detection model to be trained, and performing target detection on the sample image through the target detection model to obtain a heat map to be optimized, specifically includes:

[0030] For each sample image, input the sample image into the target detection model to be trained, and perform target detection on the sample image through the target detection model to obtain a heat map to be optimized under the spectrum corresponding to the sample image;

[0031] Taking the minimization of the deviation between the coordinate information corresponding to the image of the sample target in the sample image included in the heat map to be optimized and the label coordinates corresponding to the sample image, and the deviation between the target information corresponding to the image of the sample target in the sample image included in the heat map to be optimized and the label target information corresponding to the sample image as the optimization objective, training the target detection model, specifically includes:

[0032] For each sample image, taking the minimization of the deviation between the coordinate information corresponding to the image of the sample target in the heat map to be optimized under the spectrum corresponding to the sample image and the label coordinates corresponding to the sample image, and the deviation between the target information corresponding to the image of the sample target in the heat map to be optimized under the spectrum corresponding to the sample image and the label target information corresponding to the sample image as the optimization objective, train the target detection model.

[0033] This specification provides a target detection device, which is used to execute a target detection method. The target detection method performs target detection on a distant target through a spectral compressive imaging system. The spectral compressive imaging system includes: a telescopic imaging group, a defocusing lens group, a planar array detector, and a target detection unit. The telescopic imaging group includes: an objective lens and an eyepiece, including:

[0034] An acquisition module, configured to acquire the reflected light of the distant target as incident light;

[0035] A processing module, configured to perform aperture compression on the incident light through the objective lens and the eyepiece in the telescopic imaging group;

[0036] An imaging module, configured to process the incident light after aperture compression through the defocusing lens group to expand the imaging area of the target, and obtain a spectral compressive image with an expanded imaging area of the target through the planar array detector;

[0037] A detection module, configured to input the spectral compressive image into a pre-trained target detection model deployed in the target detection unit, and perform target detection on the spectral compressive image through the target detection model.

[0038] Optionally, the spectral compressive imaging system further includes: a diffractive optical element;

[0039] Before the imaging module processes the incident light after aperture compression through the defocusing lens group, the processing module is further configured to perform amplitude and phase modulation on different spectra corresponding to the incident light at different wavelengths through the diffractive optical element to obtain multiple spectra at different wavelengths;

[0040] The imaging module is specifically configured to process multiple spectra at different wavelengths obtained after modulation by the diffractive optical element through the defocusing lens group to expand the imaging areas of the spectra of the target at different wavelengths;

[0041] The imaging module is specifically configured to integrate and accumulate the expanded imaging areas of the spectra of the target at different wavelengths through the planar array detector to obtain the spectral compressive image of the target.

[0042] This specification provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned target detection method is implemented.

[0043] This specification provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the above-mentioned method for object detection is implemented.

[0044] The above-mentioned at least one technical solution adopted in this specification can achieve the following beneficial effects:

[0045] As can be seen from the above method, in the method for object detection provided in this specification, the reflected light of a distant object is obtained as the incident light, the incident light is processed by a telescopic imaging group and a diffractive optical element, the processed incident light is diffusely imaged by a defocus lens group, and then a spectral compression image is obtained by a area array detector. Finally, the trained object detection model preset in the object detection unit is used to perform object detection on the spectral compression image.

[0046] As can be seen from the above content, the method for object detection provided in this specification can diffusely magnify the spectral compression image of a distant object through a defocus lens group, which is convenient for object positioning of the distant object, and at the same time makes the feature information of the distant object richer, facilitating the identification of the type of the distant object and improving the accuracy of the detection result of the distant object. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The drawings described herein are used to provide a further understanding of this specification, and constitute a part of this specification. The schematic embodiments of this specification and their descriptions are used to explain this specification and do not constitute an improper limitation to this specification. In the drawings:

[0048] Figure 1 is a schematic flowchart of a method for object detection provided in this specification;

[0049] Figure 2 is a schematic diagram of the spectrum of an object at different wavelengths provided in this specification;

[0050] Figure 3 is a schematic diagram of an object imaging through a point spread function at different wavelengths provided in this specification;

[0051] Figure 4 is a schematic diagram of a spectral compression imaging system provided in this specification;

[0052] Figure 5 is a schematic diagram of a device for object detection provided in this specification;

[0053] Figure 6 is provided in this specification corresponding to Figure 1 is a schematic diagram of the structure of an electronic device. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] To make the objectives, technical solutions, and advantages of this specification clearer, the technical solutions of this specification will be clearly and completely described below in conjunction with specific embodiments of this specification and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all of them. Based on the embodiments in this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope protected by this specification.

[0055] The following will detail the technical solutions provided by each embodiment of this specification in conjunction with the drawings.

[0056] Figure 1 The following is a schematic flowchart of a method for object detection provided by this specification, including the following steps:

[0057] S101: Obtain the reflected light of a distant object as incident light.

[0058] In the field of image recognition, object detection of distant objects is relatively difficult. The imaging image of a distant object obtained during the object detection process contains too few pixel points, which causes difficulties in the object recognition process of the object detection system. Therefore, how to improve the accuracy of object detection of distant objects is particularly important.

[0059] Based on this, in this specification, the object detection device can obtain the reflected light of a distant object as incident light. Among them, due to the relatively large distance between the distant object and the object detection device, the included angle between the reflected light of the distant object and the optical reference axis of the object detection device is very small, much less than 1 rad. Therefore, the sine value and tangent value of the included angle between the reflected light and the optical reference axis of the object detection device are approximately equal, and the value is approximately equal to the included angle. It satisfies the paraxial approximation condition in optics. When the reflected light of the distant object is incident on the object detection device, there is no need to process the reflected light of the distant object, and the object detection device can directly use the reflected light of the distant object as incident light.

[0060] In this specification, the above-mentioned object detection device can be an electronic device such as a desktop computer or a notebook computer, or a computing device with optical elements. For the convenience of description, below, only the case where the object detection device is the execution subject will be used to illustrate the object detection method provided by this specification.

[0061] Based on this, the object detection device applying the object detection method provided by this specification can be used for object detection tasks of distant objects, such as object detection scenarios such as long-distance detection of wild animals.

[0062] S102: Compress the aperture of the incident light through the objective lens and the eyepiece in the telescopic imaging group.

[0063] S103: Process the incident light after aperture compression through the defocus lens group to expand the imaging area of the target object, and obtain a spectral compression image with the expanded imaging area of the target object through the area array detector.

[0064] After the target detection device acquires the incident light, it compresses the aperture of the incident light by using the telescopic imaging group. The above telescopic imaging group is composed of an objective lens and an eyepiece, and a Kepler type structure can be adopted, that is, both the objective lens and the eyepiece can be selected as convex lenses. The objective lens can be a convex lens with a curvature radius of 200 mm, a diameter of 80 mm, and a thickness of 10 mm, and the eyepiece can be a convex lens with a curvature radius of 70 mm, a diameter of 30 mm, and a thickness of 10 mm. The objective lens and the eyepiece are made of glass material, and K9 type can be uniformly adopted. The placement distance between the objective lens and the eyepiece can be 240 mm. The objective lens and the eyepiece compress the aperture of the received incident light, reducing the incident light proportionally.

[0065] Then, process the incident light after aperture compression through the defocus lens group, image the target object, expand the imaging area of the target object, and then detect and obtain the spectral compression image of the expanded target object by the area array detector.

[0066] Among them, the incident light after aperture compression by the telescopic imaging group does not directly enter the defocus lens group, and there is also a diffractive optical element between the telescopic imaging group and the defocus lens group. Before processing the incident light after aperture compression through the defocus lens group, the diffractive optical element can modulate the amplitude and phase of the incident light after aperture compression, so that the incident light can be diffracted into multiple spectra of different wavelengths. Correspondingly, the defocus lens group can process the multiple spectra of different wavelengths diffracted by the diffractive optical element for imaging operations.

[0067] Figure 2 It is a schematic diagram of the spectra of a target object at different wavelengths provided in this specification.

[0068] From Figure 2 it can be seen that the incident light after aperture compression can diffract and distinguish the spectra at different wavelengths through the diffractive optical element, and at the same time can also spectroscopically distinguish the imaging area of the target object from the background area, so as to obtain the spectral compression images of the same target object at different wavelengths.

[0069] In this specification, the diffractive optical element can be a ground glass device with a diameter of 30 mm and a thickness of 5 mm. The glass material can be of the K9 type, and the placement distance from the eyepiece in the telescopic imaging group can be 10 mm.

[0070] It should be noted that the diffractive optical element only serves to diffract the incident light to distinguish spectra of different wavelengths, and will not compress or extend the incident light after aperture compression again. Therefore, the incident light received by the defocus lens group is essentially still the incident light after aperture compression.

[0071] Furthermore, when the defocus lens group images the incident light after aperture compression processing, there is no need to adjust factors such as the optical power distribution and deflection burden of each lens group during the imaging process to completely cancel out the aberration values. That is, the defocus lens group has no strict requirements for aberration correction during the imaging process. The above method is to use the defocus lens group to generate a defocus image of the target object on the detection image plane of the area array detector, so that the same characteristic information of the target object can be diffusely distributed to multiple detection pixels of the area array detector to achieve the purpose of diffuse imaging.

[0072] In this specification, the defocus lens group can be a cylindrical lens group orthogonally composed of a set of cylindrical lenses. The cylindrical lens group includes a front cylindrical lens and a rear cylindrical lens. Among them, the curvature direction of the front cylindrical lens is along the x-axis, the curvature direction of the rear cylindrical lens is along the y-axis, the placement distance between the front cylindrical lens and the rear cylindrical lens can be 10 mm, and the placement distance between the front cylindrical lens and the diffractive optical element can be 10 mm. The cylindrical lenses in the above cylindrical lens group can be cylindrical lenses with a diameter of 30 mm, a thickness of 10 mm, and a radius of curvature of 32 mm. The glass material of the cylindrical lens can be of the K9 type.

[0073] Furthermore, when the incident light is modulated by the diffractive optical element and imaged on the detection image plane of the area array detector through the defocus lens group, the spectral points corresponding to the characteristic information of the target object in the spectral compression image on the detection image plane are not in a focused state, but in a magnified and diffused state, so that the imaging area of the target object can be diffusely imaged and can be distinguished and displayed from the background area. Therefore, the target detection device does not need to add a spectroscopic device or a spectral scanning device to process the imaging image, and the spectral compression image of the target object can be directly generated through the area array detector.

[0074] It should be noted that for the spectra of different wavelengths modulated by the diffractive optical element, different diffusion effects can also be presented on the area array detector through a preset point spread function, such as Figure 3 shown.

[0075] Figure 3 It is a schematic diagram of imaging of a target object at different wavelengths through a point spread function provided in this specification.

[0076] It can be seen from Figure 3 that for spectra of different wavelengths, through the defocus lens group, the imaging area of the target object under the spectrum of this wavelength can be diffused, forming an imaging area as shown in Figure 3 . In Figure 3 , after the imaging areas of the target object in spectra of different wavelengths are diffused by the point spread function, the obtained areas will be different.

[0077] In this specification, the area array detector can be a monochromatic CMOS detector with a resolution of 1280*1920. The diameter of the area array detector can be 10 mm, and the placement distance from the rear cylindrical lens in the defocus lens group can be 51 mm. Of course, the area array detector can also be of other specifications, and this specification does not limit the specific specifications of the area array detector.

[0078] To facilitate the description of the process of target detection by the entire spectral compression imaging system, the following will use a schematic diagram to introduce the spectral compression imaging system, as shown in Figure 4 .

[0079] Figure 4 It is a schematic diagram of a spectral compression imaging system provided by this specification.

[0080] In Figure 4 the spectral compression imaging system shown, from left to right are: an objective lens, an eyepiece, a diffractive optical element, a defocus lens group, an area array detector, and a target detection unit. After the spectral compression imaging system obtains the long-distance reflected light of a distant target object (i.e., the trees in Figure 4 ) as the incident light, the telescopic imaging group composed of the objective lens and the eyepiece compresses the aperture of the incident light, and then the incident light after aperture compression is incident on the diffractive optical element. The diffractive optical element separates the spectra of different wavelengths, and the light is incident into the defocus lens group, and the defocus lens group diffuses the distant target object, and then inputs the imaging image in the defocus diffusion state to the detection image plane of the area array detector. Among them, the imaging area with the target object feature information on the imaging image in the defocus diffusion state will be diffusely distributed on multiple pixels of the detection image plane. The area array detector integrates and accumulates the relatively blurred imaging images of different wavelengths in the defocus diffusion state to achieve spectral compression, and then obtains the spectral compression image of the target object. Finally, the target detection unit performs target detection on the spectral compression image.

[0081] It should be noted that this specification mainly focuses on target detection for distant target objects. Therefore, in practical applications, the imaging area of the distant target object on the area array detector will be relatively small in the end, often only as large as a "point". Therefore, the above Figure 4For the sake of easy understanding, an example is given with distant trees as the target object.

[0082] S104: Input the spectral compression image into a pre-trained object detection model deployed in the object detection unit, so as to perform object detection on the spectral compression image through the object detection model.

[0083] After the object detection device obtains the spectral compression image of the target object through the area array detector, the spectral compression image can be input into the object detection model in the object detection unit, and the object detection model is used to perform object detection on the spectral compression image.

[0084] Among them, the object detection model in the object detection unit is an object detection model trained through deep learning, which can generate a heat map corresponding to the target object according to the obtained spectral compression image of the target object, and output the object detection result according to the heat map.

[0085] It should be noted that the heat map generated according to the spectral compression image of the target object contains the coordinate information and target information of the target object. The coordinate information is used to represent the display position of the target object on the heat map, and the target information is used to represent the category information of the target object. Furthermore, the target information in the heat map is matched with the preset target category information in the object detection model, so as to determine the category to which the target object belongs.

[0086] Furthermore, when performing object detection, the preset object detection model in the object detection unit is an object detection model that has been trained in advance. Among them, during the model training process, the object detection model to be trained can perform object detection training on the sample images of multiple sample target objects. It should be noted that the sample images of the sample target objects can be artificially selected and determined in advance, and are specifically used for the training process. The sample images are marked with the positioning information and target information of the sample target objects, that is, the label coordinates and the label target information. The label coordinates can be used to represent the center point of the imaging area of the sample target object in the sample image, and the label target information can be used to represent the category to which the sample target object in the sample image belongs determined by humans.

[0087] After determining the sample images used for training, use the object detection model to be trained to perform object detection training on the sample images. Among them, the object detection model can specifically but not limited to adopt various neural networks. For the sake of easy explanation, the convolutional neural network is taken as an example for explanation below.

[0088] A convolutional neural network can be composed of a feature extraction network and a deconvolution network. The feature extraction network can select the ResNet network framework to extract the target feature information of the sample object in the sample image and input the target feature information into the deconvolution network composed of two deconvolution modules to obtain the heat map to be optimized through the deconvolution network. Among them, the heat map to be optimized contains the coordinate information and target information corresponding to the image of the sample object in the sample image recognized by the convolutional neural network.

[0089] Finally, with the goal of minimizing the deviation between the coordinate information corresponding to the image of the sample object in the sample image contained in the heat map to be optimized and the label coordinates corresponding to the sample image, and the deviation between the target information corresponding to the image of the sample object in the sample image contained in the heat map to be optimized and the label target information corresponding to the sample image, the object detection model is trained.

[0090] Among them, the first loss can be determined by the deviation between the coordinate information corresponding to the image of the sample object in the sample image contained in the heat map to be optimized and the label coordinates corresponding to the sample image, and the second loss can be determined by the deviation between the target information corresponding to the image of the sample object in the sample image contained in the heat map to be optimized and the label target information corresponding to the sample image. Then, with the goal of minimizing the sum of the first loss and the second loss, the object detection model is trained.

[0091] It should be noted that in order to improve the object detection ability of the object detection model for objects with spectra of different wavelengths, make the detectable wavelength range of the object detection model wider, and thus make the object detection model can detect objects more accurately. During the training process, the diffraction function of the diffractive optical element can be used to enable the object detection model to be trained to detect multiple sample images corresponding to the spectra of the same sample object with different wavelengths, and then optimize the heat maps of different wavelengths generated by the object detection model to be trained, so as to optimize and iterate the object detection model to achieve the purpose of training the object detection model to be trained.

[0092] Specifically, for each sample image, the object detection device can input the sample image into the object detection model to be trained, so as to detect the sample image through the object detection model and obtain the heat map to be optimized under the corresponding spectrum of the sample image.

[0093] Subsequently, for each sample image, the target detection device can minimize the deviation between the coordinate information corresponding to the image of the sample target object included in the heat map to be optimized under the spectrum corresponding to the sample image and the label coordinates corresponding to the sample image, as well as the deviation between the target information corresponding to the image of the sample target object in the sample image included in the heat map to be optimized under the spectrum corresponding to the sample image and the label target information corresponding to the sample image, and use this as the optimization objective to train the target detection model.

[0094] Among them, since there are multiple sample images, and according to the heat map to be optimized corresponding to each sample image and the label information, corresponding loss values can be determined. Then, the loss values determined by the deviation between the heat map to be optimized corresponding to each sample image and the label information can be summed up. Furthermore, using the minimization of the overall loss value as the optimization objective to train the target detection model, so that the trained target detection model can effectively detect spectra of different wavelengths.

[0095] As can be seen from the above content, the target detection method provided in this specification can diffusely amplify the spectral compression image of a distant target object through a defocus lens group, which facilitates target positioning of the distant target object while making the feature information of the distant target object more obvious, facilitating the recognition of the type of the distant target object, and improving the accuracy of the detection result of the distant target object.

[0096] The above is the method of one or more embodiments of this specification. Based on the same idea, this specification also provides a corresponding target detection device, as Figure 5 shown.

[0097] Figure 5 FIG. is a schematic diagram of a target detection device provided in this specification, including:

[0098] An acquisition module 501: configured to acquire the reflected light of a distant target object as incident light;

[0099] A processing module 502: configured to perform aperture compression on the incident light through the objective lens and the eyepiece in the telescopic imaging group;

[0100] An imaging module 503: configured to process the incident light after aperture compression through the defocus lens group to expand the imaging area of the target object, and obtain a spectral compression image of the expanded imaging area of the target object through the area array detector;

[0101] A detection module 504: configured to input the spectral compression image into a pre-trained target detection model deployed in the target detection unit, and perform target detection on the spectral compression image through the target detection model.

[0102] Optionally, the spectral compressive imaging system further includes: a diffractive optical element;

[0103] Before the imaging module 503 processes the incident light after aperture compression through the defocus lens group, the processing module 502 is further configured to perform amplitude and phase modulation on different spectra corresponding to the incident light at different wavelengths through the diffractive optical element, so as to obtain multiple spectra at different wavelengths;

[0104] The imaging module 503 is specifically configured to process multiple spectra at different wavelengths obtained after modulation by the diffractive optical element through the defocus lens group, so as to expand the imaging regions of the spectra of the target object at different wavelengths;

[0105] The imaging module 503 is specifically configured to integrate and accumulate the expanded imaging regions of the spectra of the target object at different wavelengths through the area array detector, so as to obtain the spectral compressive image of the target object.

[0106] Optionally, the detection module 504 is specifically configured to input the spectral compressive image into a pre-trained target detection model deployed in the target detection unit, so as to output, through the target detection model, a heat map for the spectral compressive image, where the heat map is used to represent the coordinate information of the target object and the target information of the target object; and obtain a target detection result according to the heat map.

[0107] Optionally, the device further includes:

[0108] A training module 505, configured to obtain a sample image; input the sample image into a target detection model to be trained, so as to perform target detection on the sample image through the target detection model to obtain an unoptimized heat map to be optimized; and use minimizing the deviation between the coordinate information corresponding to the image of the sample target object in the sample image included in the unoptimized heat map to be optimized and the label coordinates corresponding to the sample image, and the deviation between the target information corresponding to the image of the sample target object in the sample image included in the unoptimized heat map to be optimized and the label target information corresponding to the sample image as an optimization objective to train the target detection model.

[0109] Optionally, the training module 505 is specifically configured to obtain each sample image of the same sample target object, where different sample images correspond to spectra of different wavelengths;

[0110] The training module 505 is specifically configured to, for each sample image, input the sample image into a target detection model to be trained, so as to perform target detection on the sample image through the target detection model to obtain an unoptimized heat map to be optimized corresponding to the spectrum of the sample image;

[0111] Specifically, for each sample image, the training module 505 trains the object detection model with the optimization objective of minimizing the deviation between the coordinate information corresponding to the image of the sample object included in the heat map to be optimized under the spectrum corresponding to the sample image and the label coordinates corresponding to the sample image, and minimizing the deviation between the target information corresponding to the image of the sample object in the sample image included in the heat map to be optimized under the spectrum corresponding to the sample image and the label target information corresponding to the sample image.

[0112] This specification also provides a computer-readable storage medium storing a computer program that can be used to execute the Figure 1 object detection method provided above.

[0113] This specification also provides Figure 6 a schematic structural diagram of an electronic device corresponding to Figure 1 as shown. At the hardware level, as Figure 6 shown, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory. Of course, it may also include other hardware required for other services. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to implement the Figure 1 object detection method described above.

[0114] In the 1990s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to circuit structures such as diodes, transistors, switches, etc.) or software improvements (improvements to method flows). However, with the development of technology, many method flow improvements today can be regarded as direct improvements to hardware circuit structures. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that an improvement to a method flow cannot be implemented using a hardware entity module. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logical function is determined by the user programming the device. Designers can program themselves to "integrate" a digital system onto a single PLD, without having to ask a chip manufacturer to design and fabricate a dedicated integrated circuit chip. Moreover, today, instead of manually fabricating integrated circuit chips, this programming is mostly implemented using "logic compiler" software, which is similar to the software compilers used in program development and writing. The original code before compilation also has to be written in a specific programming language, which is called a Hardware Description Language (HDL), and there is not just one type of HDL, but many types, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones currently are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also be aware that by simply performing a little logical programming on the method flow using the above-mentioned several hardware description languages and programming it into an integrated circuit, it is easy to obtain the hardware circuit that implements the logical method flow.

[0115] The controller can be implemented in any suitable manner. For example, the controller can take the form of, for example, a microprocessor or a processor and a computer-readable medium storing computer-readable program code (such as software or firmware) executable by the (micro)processor, logic gates, switches, an application specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of the controller include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in the form of pure computer-readable program code, it is entirely possible to logically program the method steps to enable the controller to be implemented in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, and embedded microcontrollers to achieve the same function. Therefore, such a controller can be considered a hardware component, and the devices included therein for implementing various functions can also be regarded as structures within the hardware component. Or even, the devices for implementing various functions can be regarded as either software modules for implementing the method or structures within the hardware component.

[0116] The systems, devices, modules, or units illustrated in the above embodiments can be specifically implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0117] For the convenience of description, when describing the above devices, they are described separately as various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0118] Those skilled in the art should understand that the embodiments of this specification can be provided as a method, a system, or a computer program product. Therefore, this specification can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program code.

[0119] This specification is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the specification. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, such that the instructions executed by the processor of the computer or other programmable data processing device generate means for implementing the specified functions in the process Figure 1 one process or multiple processes and / or blocks Figure 1 or means for implementing the specified functions in multiple blocks.

[0120] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including instruction means that implement the specified functions in the process Figure 1 one process or multiple processes and / or blocks Figure 1 or means for implementing the specified functions in multiple blocks.

[0121] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the specified functions in the process Figure 1 one process or multiple processes and / or blocks Figure 1 or means for implementing the specified functions in multiple blocks.

[0122] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and memory.

[0123] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM), and / or non-volatile memory such as read-only memory (ROM) or flash memory (flash RAM). The memory is an example of computer-readable media.

[0124] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0125] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0126] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Thus, this specification may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0127] This specification may be described in the general context of computer-executable instructions, such as program modules, executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media, including storage devices.

[0128] The various embodiments in this specification are described in a progressive manner. For the same or similar parts among the various embodiments, reference can be made to each other, and the key points of each embodiment are the differences from other embodiments. In particular, for the system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and for the relevant parts, reference can be made to the partial description of the method embodiments.

[0129] The above description is only for the embodiments of this specification and is not intended to limit this specification. For those skilled in the art, various modifications and changes can be made to this specification. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of this specification shall be included within the scope of the claims of this specification.

Claims

1. A target detection method, characterized in that, The target detection method performs target detection on a distant target object through a spectral compressive imaging system, and the spectral compressive imaging system includes: a telescopic imaging group, a defocus lens group, a matrix detector, and a target detection unit. The telescopic imaging group includes: an objective lens and an eyepiece. The method includes: Obtain the reflected light of the distant target object as incident light; Perform aperture compression on the incident light through the objective lens and the eyepiece in the telescopic imaging group; Process the incident light after aperture compression through the defocus lens group to expand the imaging area of the target object, and obtain a spectral compressive image with an expanded imaging area of the target object through the matrix detector; Input the spectral compressive image into a pre-trained target detection model deployed in the target detection unit to perform target detection on the spectral compressive image through the target detection model.

2. The method according to claim 1, characterized in that, The spectral compressive imaging system further includes: a diffractive optical element; Before processing the incident light after aperture compression through the defocus lens group, the method further includes: Perform amplitude and phase modulation on different spectra corresponding to the incident light at different wavelengths through the diffractive optical element to obtain multiple spectra at different wavelengths; Processing the incident light after aperture compression through the defocus lens group to expand the imaging area of the target object specifically includes: Process multiple spectra at different wavelengths obtained after modulation by the diffractive optical element through the defocus lens group to expand the imaging areas of the spectra of the target object at different wavelengths; Obtaining a spectral compressive image with an expanded imaging area of the target object through the matrix detector specifically includes: Integrate and accumulate the expanded imaging areas of the spectra of the target object at different wavelengths through the matrix detector to obtain the spectral compressive image of the target object.

3. The method according to claim 1, characterized in that, Processing the incident light after aperture compression through the defocus lens group to expand the imaging area of the target object specifically includes: Perform defocus processing on the incident light after aperture compression through the defocus lens group to distinguish the imaging area of the target object from the background and expand the imaging area of the target object.

4. The method according to claim 1, characterized in that, Inputting the spectral compressive image into a pre-trained target detection model deployed in the target detection unit to perform target detection on the spectral compressive image through the target detection model specifically includes: Input the spectral compressive image into a pre-trained target detection model deployed in the target detection unit to output a heat map for the spectral compressive image through the target detection model, and the heat map is used to represent the coordinate information of the target object and the target information of the target object; Obtain the target detection result according to the heat map.

5. The method according to claim 4, wherein Training the target detection model specifically includes: Obtain a sample image; Input the sample image into the target detection model to be trained to perform target detection on the sample image through the target detection model to obtain a heat map to be optimized; Taking the minimization of the deviation between the coordinate information corresponding to the image of the sample target in the sample image included in the heat map to be optimized and the label coordinates corresponding to the sample image, and the deviation between the target information corresponding to the image of the sample target in the sample image included in the heat map to be optimized and the label target information corresponding to the sample image as the optimization objective, train the object detection model.

6. The method according to claim 5, characterized in that, Obtain sample images, specifically including: Obtain each sample image for the same sample target, where different sample images correspond to spectra of different wavelengths; Input the sample images into the object detection model to be trained, so as to perform object detection on the sample images through the object detection model to obtain a heat map to be optimized, specifically including: For each sample image, input the sample image into the object detection model to be trained, so as to perform object detection on the sample image through the object detection model to obtain the heat map to be optimized under the spectrum corresponding to the sample image; Taking the minimization of the deviation between the coordinate information corresponding to the image of the sample target in the sample image included in the heat map to be optimized and the label coordinates corresponding to the sample image, and the deviation between the target information corresponding to the image of the sample target in the sample image included in the heat map to be optimized and the label target information corresponding to the sample image as the optimization objective, train the object detection model, specifically including: For each sample image, taking the minimization of the deviation between the coordinate information corresponding to the image of the sample target included in the heat map to be optimized under the spectrum corresponding to the sample image and the label coordinates corresponding to the sample image, and the deviation between the target information corresponding to the image of the sample target included in the heat map to be optimized under the spectrum corresponding to the sample image and the label target information corresponding to the sample image as the optimization objective, train the object detection model.

7. A device for object detection, characterized in that, The device is used to execute an object detection method, and the object detection method performs object detection on a distant target through a spectral compressive imaging system, and the spectral compressive imaging system includes: a telescopic imaging group, a defocus lens group, a area array detector, and an object detection unit. The telescopic imaging group includes: an objective lens and an eyepiece, including: An acquisition module, configured to acquire the reflected light of the distant target as incident light; A processing module, configured to perform aperture compression on the incident light through the objective lens and the eyepiece in the telescopic imaging group; An imaging module, configured to process the incident light after aperture compression through the defocus lens group to expand the imaging area of the target, and obtain a spectral compressive image with the imaging area of the target expanded through the area array detector; A detection module, configured to input the spectral compressive image into a pre-trained object detection model deployed in the object detection unit, so as to perform object detection on the spectral compressive image through the object detection model.

8. The device according to claim 7, characterized in that, The spectral compressive imaging system further includes: a diffractive optical element; Before the imaging module processes the incident light after aperture compression through the defocus lens group, the processing module is further configured to perform amplitude and phase modulation on different spectra corresponding to the incident light at different wavelengths through the diffractive optical element to obtain multiple spectra of different wavelengths; Specifically, the imaging module is configured to process the multiple spectra of different wavelengths obtained after modulation by the diffractive optical element through the defocus lens group to expand the imaging regions of the spectra of the target object at different wavelengths; Specifically, the imaging module is configured to integrate and accumulate the expanded imaging regions of the spectra of the target object at different wavelengths through the area array detector to obtain a spectral compression image of the target object.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, and when the computer program is executed by a processor, the method described in any one of claims 1 to 6 above is implemented.

10. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, the method described in any one of claims 1 to 6 above is implemented.

Images