Three-dimensional imaging method and device of compound eye camera based on focal length pyramid network

Abstract

The present invention discloses a three-dimensional imaging method and device of a compound eye camera based on a focal length pyramid network, the method comprising: determining the device parameters corresponding to the target acquisition device for acquiring depth images and the acquisition control parameters of all target acquisition devices according to the determined focal length pyramid network, the device parameters including the even number of devices, the image acquisition parameters of each target acquisition device, the acquisition control parameters including the layout parameters and the acquisition direction; performing device parameter configuration and acquisition deployment operations on all target acquisition devices according to the device parameters and the acquisition control parameters, and obtaining the target depth images acquired by all target acquisition devices; and then performing deep processing operations on all target depth images to obtain the deep processing results corresponding to all target depth images. It can be seen that the implementation of the present invention can improve the processing accuracy of the depth images and improve the image quality of the obtained depth images.

Description

Technical Field

[0001] The present invention relates to the field of image depth processing technology, and in particular to a three-dimensional imaging method and device of a compound eye camera based on a focal pyramid network. Background Art

[0002] With the development of image processing technology, after image processing technology was put into production application, the technical requirements for image processing in various industries have also increased accordingly. Among them, the application of image three-dimensional technology is the top priority. For example, in industrial production and manufacturing, augmented reality, object modeling, entertainment games and many other fields, the acquisition of depth image information is involved.

[0003] Based on the above-mentioned status quo, it is easy to think that in addition to the acquisition of depth image information, the analysis, processing and optimization of depth image information are all directions that need to be improved nowadays. However, the processing algorithms currently used to process depth images will eventually obtain depth images with, for example: binocular parallax occlusion areas, weak texture areas, repeated texture areas, etc., that is, the depth image processing effect is not good. It can be seen that how to improve the processing accuracy of depth images and improve the image quality of the obtained depth images is particularly important. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a three-dimensional imaging method and device of a compound eye camera based on a focal pyramid network, which can improve the processing accuracy of depth images, and it is particularly important to improve the image quality of the obtained depth images.

[0005] In order to solve the above technical problems, the first aspect of the present invention discloses a three-dimensional imaging method of a compound eye camera based on a focal pyramid network, the method comprising:

[0006] Determine, according to the determined focal length pyramid network, device parameters corresponding to the target acquisition device for acquiring the depth image, wherein the device parameters include the number of devices and image acquisition parameters corresponding to each of the target acquisition devices, wherein the number of devices is an even number;

[0007] Determine acquisition control parameters for all target acquisition devices according to the focal length pyramid network, wherein the acquisition control parameters include layout parameters and acquisition directions, wherein the layout parameters are used to deploy image acquisition positions of each target acquisition device, and the acquisition directions include vertical acquisition directions and / or horizontal acquisition directions;

[0008] According to the device parameters and the acquisition control parameters, perform device parameter configuration operations and acquisition deployment operations on all the target acquisition devices, and obtain target depth images to be processed acquired by all the target acquisition devices, where all the target depth images are images for the same acquisition target;

[0009] According to the focal length pyramid network, a deep processing operation is performed on all the target depth images to obtain deep processing results corresponding to all the target depth images, wherein the deep processing operation includes a motion information optimization operation.

[0010] As an optional implementation, in the first aspect of the present invention, the image acquisition parameters corresponding to each of the target acquisition devices include a resolution and an image acquisition frame rate corresponding to the target acquisition device:

[0011] The focal pyramid network includes multiple network levels, and the collection control parameters for all the target collection devices also include the number of network layers, wherein each of the network levels is deployed with a number of target collection devices that matches the network level, the number of target collection devices between adjacent network levels is a preset multiple relationship, and the number of devices in the lower network level is less than the number of devices in the higher network level.

[0012] As an optional implementation, in the first aspect of the present invention, in the focal length pyramid network, a preset parallel computing mode is adopted for the target depth images at different network levels;

[0013] In the same network level of the focal pyramid network, there is at least one corresponding alignment acquisition device in the network level for each of the target acquisition devices, and the target acquisition device and all the alignment acquisition devices corresponding to the target acquisition device use the parallel computing mode as the operation mode for processing the target depth images respectively acquired.

[0014] As an optional implementation, in the first aspect of the present invention, the device type of the target acquisition device includes a main camera type and a secondary camera type, and the target acquisition devices corresponding to different device types perform different depth image acquisition operations; the device parameter configuration operation and the acquisition deployment operation are performed on all the target acquisition devices according to the device parameters and the acquisition control parameters, including:

[0015] According to the number of devices, determining the number of levels of the target acquisition devices required to be deployed at each network level in the focal pyramid network;

[0016] According to the number of levels of each of the network levels, determine the number of main cameras of the target acquisition devices of the main camera type and the number of secondary cameras of the target acquisition devices of the secondary camera type in the network level;

[0017] For each of the network levels, taking the acquisition direction as the reference deployment direction, performing acquisition deployment operations according to the number of main cameras and the number of secondary cameras corresponding to the network level and the image acquisition position of each of the target acquisition devices in the network level, so that the deployment of each of the target acquisition devices in the network level meets the deployment requirements corresponding to the network level;

[0018] After determining that all the target acquisition devices in each of the network levels have completed the acquisition deployment operation, a device parameter configuration operation is performed on the target acquisition device according to the device parameters of each target acquisition device.

[0019] As an optional implementation, in the first aspect of the present invention, performing a deep processing operation on all the target depth images according to the focal pyramid network to obtain deep processing results corresponding to all the target depth images includes:

[0020] According to the focal length pyramid network, an initial depth calculation operation is performed on all the target depth images to obtain initial depth calculation results corresponding to all the target depth images;

[0021] For the target depth image acquired by each of the target acquisition devices, a motion information optimization operation is performed on the initial depth calculation result corresponding to the target depth image according to the determined motion information optimization algorithm to obtain a motion information optimized image corresponding to the initial depth calculation result corresponding to the target depth image;

[0022] According to the progressive optimization algorithm corresponding to the focal length pyramid network, it is determined whether there is an image to be corrected in all the motion information optimized images, wherein the image to be corrected is an image corresponding to an area in a certain motion information optimized image that matches a preset correction area;

[0023] When it is determined that the image to be corrected exists in all the motion information optimized images, a deep correction optimization operation is performed on the image to be corrected according to the regional correction algorithm and the correction type corresponding to the image to be corrected, so that the image to be corrected is adjusted and optimized to meet the preset deep processing requirements;

[0024] When it is determined that the image to be corrected does not exist in all the motion information optimized images, all the motion optimized images are determined to be images that meet the preset deep processing requirements as the deep processing results corresponding to the target depth image corresponding to the target acquisition device.

[0025] As an optional implementation, in the first aspect of the present invention, for each target acquisition device, the target depth image acquired, performing a motion information optimization operation on the initial depth calculation result corresponding to the target depth image according to the determined motion information optimization algorithm, and obtaining a motion information optimized image corresponding to the initial depth calculation result corresponding to the target depth image, including:

[0026] For the target depth image acquired by each of the target acquisition devices, a motion information analysis operation is performed on the target depth image to obtain motion information between adjacent frame images in the target depth image, and a feature matching optimization operation is performed according to the motion information and the alignment motion information to obtain a feature matching optimization result corresponding to the motion information and the alignment motion information as a motion information optimized image corresponding to the target acquisition device;

[0027] The alignment motion information is information obtained after the alignment device performs the motion information analysis operation on the target depth image corresponding to the alignment device, and the alignment device is the alignment acquisition device corresponding to the acquisition device.

[0028] As an optional implementation, in the first aspect of the present invention, the type to be corrected corresponding to the image to be corrected includes a binocular parallax occlusion type, a weak texture type or a repeated texture type;

[0029] Among them, in the focal length pyramid network, the image acquisition parameters corresponding to all the target acquisition devices of each network level and all the target acquisition devices of the network level adjacent to the network level respectively meet the preset multiple conditions, wherein the multiple conditions are specifically:

[0030] The quotient of the resolution corresponding to the target acquisition device in the first level and the resolution corresponding to the target acquisition device in the second level is greater than a first preset value, wherein the network level corresponding to the first level is higher than the network level corresponding to the second level, and the first level is adjacent to the second level, and the first preset value is a positive number and less than 1;

[0031] The quotient of the image acquisition frame rate corresponding to the target acquisition device in the first level and the image acquisition frame rate corresponding to the target acquisition device in the second level is equal to a second preset value, and the second preset value is a positive number and greater than 1.

[0032] The second aspect of the present invention discloses a three-dimensional imaging device of a compound eye camera based on a focal length pyramid network, the device comprising:

[0033] A determination module, used to determine device parameters corresponding to a target acquisition device for acquiring a depth image according to the determined focal length pyramid network, wherein the device parameters include the number of devices and image acquisition parameters corresponding to each of the target acquisition devices, wherein the number of devices is an even number;

[0034] The determination module is further used to determine the acquisition control parameters for all the target acquisition devices according to the focal length pyramid network, the acquisition control parameters including layout parameters and acquisition directions, the layout parameters are used to deploy the image acquisition position of each of the target acquisition devices, and the acquisition direction includes a vertical acquisition direction and / or a horizontal acquisition direction;

[0035] A deployment control module, used to perform device parameter configuration operations and collection deployment operations on all the target collection devices according to the device parameters and the collection control parameters;

[0036] An acquisition module, used to acquire target depth images to be processed acquired by all the target acquisition devices, wherein all the target depth images are images for the same acquisition target;

[0037] A deep processing module is used to perform deep processing operations on all the target depth images according to the focal length pyramid network to obtain deep processing results corresponding to all the target depth images, wherein the deep processing operations include motion information optimization operations.

[0038] As an optional implementation, in the second aspect of the present invention, the image acquisition parameters corresponding to each of the target acquisition devices include a resolution and an image acquisition frame rate corresponding to the target acquisition device:

[0039] The focal pyramid network includes multiple network levels, and the collection control parameters for all the target collection devices also include the number of network layers, wherein each of the network levels is deployed with a number of target collection devices that matches the network level, the number of target collection devices between adjacent network levels is a preset multiple relationship, and the number of devices in the lower network level is less than the number of devices in the higher network level.

[0040] As an optional implementation, in the second aspect of the present invention, in the focal pyramid network, a preset parallel computing mode is adopted for the target depth images at different network levels;

[0041] In the same network level of the focal pyramid network, there is at least one corresponding alignment acquisition device in the network level for each of the target acquisition devices, and the target acquisition device and all the alignment acquisition devices corresponding to the target acquisition device use the parallel computing mode as the operation mode for processing the target depth images respectively acquired.

[0042] As an optional implementation, in the second aspect of the present invention, the device type of the target acquisition device includes a main camera type and a secondary camera type, and the target acquisition devices corresponding to different device types perform different depth image acquisition operations; the deployment control module performs device parameter configuration operations and acquisition deployment operations on all the target acquisition devices according to the device parameters and the acquisition control parameters, specifically including:

[0043] According to the number of devices, determining the number of levels of the target acquisition devices required to be deployed at each network level in the focal pyramid network;

[0044] According to the number of levels of each of the network levels, determine the number of main cameras of the target acquisition devices of the main camera type and the number of secondary cameras of the target acquisition devices of the secondary camera type in the network level;

[0045] For each of the network levels, taking the acquisition direction as the reference deployment direction, performing acquisition deployment operations according to the number of main cameras and the number of secondary cameras corresponding to the network level and the image acquisition position of each of the target acquisition devices in the network level, so that the deployment of each of the target acquisition devices in the network level meets the deployment requirements corresponding to the network level;

[0046] After determining that all the target acquisition devices in each of the network levels have completed the acquisition deployment operation, a device parameter configuration operation is performed on the target acquisition device according to the device parameters of each target acquisition device.

[0047] As an optional implementation, in the second aspect of the present invention, the deep processing module performs a deep processing operation on all the target depth images according to the focal pyramid network, and a method of obtaining deep processing results corresponding to all the target depth images specifically includes:

[0048] According to the focal length pyramid network, an initial depth calculation operation is performed on all the target depth images to obtain initial depth calculation results corresponding to all the target depth images;

[0049] For the target depth image acquired by each of the target acquisition devices, a motion information optimization operation is performed on the initial depth calculation result corresponding to the target depth image according to the determined motion information optimization algorithm to obtain a motion information optimized image corresponding to the initial depth calculation result corresponding to the target depth image;

[0050] According to the progressive optimization algorithm corresponding to the focal length pyramid network, it is determined whether there is an image to be corrected in all the motion information optimized images, wherein the image to be corrected is an image corresponding to an area in a certain motion information optimized image that matches a preset correction area;

[0051] When it is determined that the image to be corrected exists in all the motion information optimized images, a deep correction optimization operation is performed on the image to be corrected according to the regional correction algorithm and the correction type corresponding to the image to be corrected, so that the image to be corrected is adjusted and optimized to meet the preset deep processing requirements;

[0052] When it is determined that the image to be corrected does not exist in all the motion information optimized images, all the motion optimized images are determined to be images that meet the preset deep processing requirements as the deep processing results corresponding to the target depth image corresponding to the target acquisition device.

[0053] As an optional implementation, in the second aspect of the present invention, the deep processing module performs a motion information optimization operation on the initial depth calculation result corresponding to the target depth image collected by each of the target acquisition devices according to the determined motion information optimization algorithm, and obtains a motion information optimized image corresponding to the initial depth calculation result corresponding to the target depth image, specifically including:

[0054] For the target depth image acquired by each of the target acquisition devices, a motion information analysis operation is performed on the target depth image to obtain motion information between adjacent frame images in the target depth image, and a feature matching optimization operation is performed according to the motion information and the alignment motion information to obtain a feature matching optimization result corresponding to the motion information and the alignment motion information as a motion information optimized image corresponding to the target acquisition device;

[0055] The alignment motion information is information obtained after the alignment device performs the motion information analysis operation on the target depth image corresponding to the alignment device, and the alignment device is the alignment acquisition device corresponding to the acquisition device.

[0056] As an optional implementation, in the second aspect of the present invention, the type to be corrected corresponding to the image to be corrected includes a binocular parallax occlusion type, a weak texture type or a repeated texture type;

[0057] Among them, in the focal length pyramid network, the image acquisition parameters corresponding to all the target acquisition devices of each network level and all the target acquisition devices of the network level adjacent to the network level respectively meet the preset multiple conditions, wherein the multiple conditions are specifically:

[0058] The quotient of the resolution corresponding to the target acquisition device in the first level and the resolution corresponding to the target acquisition device in the second level is greater than a first preset value, wherein the network level corresponding to the first level is higher than the network level corresponding to the second level, and the first level is adjacent to the second level, and the first preset value is a positive number and less than 1;

[0059] The quotient of the image acquisition frame rate corresponding to the target acquisition device in the first level and the image acquisition frame rate corresponding to the target acquisition device in the second level is equal to a second preset value, and the second preset value is a positive number and greater than 1.

[0060] The third aspect of the present invention discloses another compound eye camera three-dimensional imaging device based on a focal length pyramid network, the device comprising:

[0061] A memory storing executable program code;

[0062] a processor coupled to the memory;

[0063] The processor calls the executable program code stored in the memory to execute the three-dimensional imaging method of the compound eye camera based on the focal pyramid network disclosed in the first aspect of the present invention.

[0064] The fourth aspect of the present invention discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute the three-dimensional imaging method of the compound eye camera based on the focal pyramid network disclosed in the first aspect of the present invention.

[0065] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0066] In an embodiment of the present invention, a three-dimensional imaging method of a compound eye camera based on a focal pyramid network is provided, the method comprising: determining device parameters corresponding to a target acquisition device for acquiring depth images according to a determined focal pyramid network, the device parameters comprising the number of devices and image acquisition parameters corresponding to each target acquisition device, wherein the number of devices is an even number; determining acquisition control parameters for all target acquisition devices according to the focal pyramid network, the acquisition control parameters comprising layout parameters and acquisition directions, the layout parameters being used to deploy image acquisition positions for each target acquisition device, the acquisition directions comprising vertical acquisition directions and / or horizontal acquisition directions; performing device parameter configuration operations and acquisition deployment operations on all target acquisition devices according to the device parameters and the acquisition control parameters, and obtaining target depth images to be processed acquired by all target acquisition devices, wherein all target depth images are images for the same acquisition target; performing deep processing operations on all target depth images according to the focal pyramid network, and obtaining deep processing results corresponding to all target depth images, wherein the deep processing operations comprise motion information optimization operations. It can be seen that the implementation of the present invention can configure and deploy each acquisition device based on the pyramid network, and then use the pyramid network to perform deep processing on the depth image collected by each acquisition device, thereby realizing the correction, supplementation and optimization of the depth image, which is beneficial to reducing the proportion of blurred areas in the depth image, improving the image quality of the final depth image, and also improving the accuracy and reliability of the depth image. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0068] Figure 1 It is a flow chart of a three-dimensional imaging method of a compound eye camera based on a focal pyramid network disclosed in an embodiment of the present invention;

[0069] Figure 2 It is a flow chart of another method for three-dimensional imaging of a compound eye camera based on a focal pyramid network disclosed in an embodiment of the present invention;

[0070] Figure 3 It is a structural schematic diagram of a three-dimensional imaging device of a compound eye camera based on a focal pyramid network disclosed in an embodiment of the present invention;

[0071] Figure 4 It is a structural schematic diagram of another compound eye camera three-dimensional imaging device based on a focal pyramid network disclosed in an embodiment of the present invention;

[0072] Figure 5 It is a schematic structural diagram of a compound eye three-dimensional imaging system including a three-layer pyramid based on a vertical collection direction disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0073] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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 only 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0074] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or end including a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units inherent to these processes, methods, products or ends.

[0075] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0076] The present invention discloses a three-dimensional imaging method and device of a compound eye camera based on a focal length pyramid network, which can configure and deploy each acquisition device based on the pyramid network, and then use the pyramid network to perform deep processing on the depth image collected by each acquisition device, thereby realizing the correction, supplementation and optimization of the depth image, which is conducive to reducing the proportion of blurred areas in the depth image, improving the image quality of the final depth image, and also improving the accuracy and reliability of the obtained depth image. The following are detailed descriptions.

[0077] Embodiment 1

[0078] See also Figure 1 , Figure 1 1 is a flow chart of a three-dimensional imaging method of a compound eye camera based on a focal pyramid network disclosed in an embodiment of the present invention. Figure 1The described three-dimensional imaging method of a compound eye camera based on a focal length pyramid network can be applied to a three-dimensional imaging device of a compound eye camera based on a focal length pyramid network, and the embodiments of the present invention are not limited thereto. Figure 1 As shown, the compound eye camera three-dimensional imaging method based on the focal length pyramid network may include the following operations:

[0079] 101. Determine device parameters corresponding to a target acquisition device for acquiring depth images based on the determined focal length pyramid network.

[0080] In an embodiment of the present invention, the device parameters include the number of devices and image acquisition parameters corresponding to each target acquisition device, where the number of devices is an even number; wherein the target acquisition device can specifically be a binocular camera configured according to preset configuration requirements, and each binocular camera uses a computer stereo vision imaging method to acquire a depth image.

[0081] In the embodiment of the present invention, the image acquisition parameters corresponding to each target acquisition device include the resolution and image acquisition frame rate corresponding to the target acquisition device:

[0082] Among them, the focal pyramid network includes multiple network levels, and the collection control parameters for all target collection devices also include the number of network layers, wherein each network level is deployed with a number of target collection devices that matches the network level, and the number of target collection devices between adjacent network levels is a preset multiple relationship, and the number of devices in the low network level is less than the number of devices in the high network level.

[0083] In an embodiment of the present invention, in a focal pyramid network, a preset parallel computing mode is adopted for target depth images at different network levels;

[0084] In the same network level of the focal pyramid network, each target acquisition device has at least one corresponding alignment acquisition device in the network level, and the target acquisition device and all the alignment acquisition devices corresponding to the target acquisition device use a parallel computing mode when processing the target depth images collected by each of them.

[0085] 102. Based on the focal pyramid network, determine the collection control parameters for all target collection devices.

[0086] In an embodiment of the present invention, the acquisition control parameters include layout parameters and acquisition directions. The layout parameters are used to deploy the image acquisition position of each target acquisition device. The acquisition direction includes a vertical acquisition direction and / or a horizontal acquisition direction.

[0087] See also Figure 5 , Figure 5 The compound eye three-dimensional imaging system including three layers of pyramids based on the vertical acquisition direction disclosed in an embodiment of the present invention, wherein:

[0088] The first layer contains 2 1 That is, 2 cameras with the shortest focal length, and the second layer contains 2 2 That is, 4 cameras with longer focal lengths, and the third layer contains 2 3 That is, 8 cameras with longer focal lengths. Similarly, the nth layer (n≥1, the same below) requires 2 n The ellipsis indicates that when more levels are needed, the number of cameras at different levels can be increased in sequence.

[0089] The algorithms in the embodiments of the present invention are described with the left camera as the primary camera and the right camera as the secondary camera. The algorithms are also applicable to the case where the right camera is the primary camera and the left camera is the secondary camera.

[0090] 103. According to the device parameters and acquisition control parameters, perform device parameter configuration operations and acquisition deployment operations on all target acquisition devices, and obtain the target depth images to be processed collected by all target acquisition devices.

[0091] In the embodiment of the present invention, all target depth images are images for the same acquisition target.

[0092] 104. According to the focal pyramid network, deep processing operations are performed on all target depth images to obtain deep processing results corresponding to all target depth images.

[0093] In the embodiment of the present invention, the deep processing operation includes a motion information optimization operation.

[0094] It can be seen that implementation Figure 1 The described three-dimensional imaging method of the compound eye camera based on the focal length pyramid network can configure and deploy each acquisition device based on the pyramid network, and then use the pyramid network to perform deep processing on the depth image collected by each acquisition device, thereby realizing the correction, supplementation and optimization of the depth image, which is beneficial to reduce the proportion of blurred areas in the depth image, improve the image quality of the final depth image, and also improve the accuracy and reliability of the obtained depth image.

[0095] In an optional embodiment, the device type of the target acquisition device includes a main camera type and a secondary camera type, and the target acquisition devices corresponding to different device types perform different depth image acquisition operations; according to the device parameters and acquisition control parameters, the method of performing device parameter configuration operations and acquisition deployment operations on all target acquisition devices specifically includes:

[0096] According to the number of devices, determine the number of target acquisition devices that need to be deployed at each network level in the focal pyramid network;

[0097] According to the number of levels of each network level, determine the number of main cameras of the main camera type target acquisition devices and the number of secondary cameras of the secondary camera type target acquisition devices in the network level;

[0098] For each network level, the acquisition direction is used as the reference deployment direction. According to the number of main cameras and secondary cameras corresponding to the network level, the acquisition deployment operation is performed according to the image acquisition position of each target acquisition device in the network level, so that the deployment of each target acquisition device in the network level meets the deployment requirements corresponding to the network level.

[0099] After determining that all target collection devices in each network layer have completed the collection deployment operation, the device parameter configuration operation is performed on the target collection device according to the device parameters of each target collection device.

[0100] In this optional embodiment, it is further described as follows: Figure 5 The numbers in the figure represent different camera numbers. Cameras 4 and 11 are the first-layer cameras of the pyramid. The distance between the two cameras is the base distance L0. In the case of an n-layer pyramid, the base distance is evenly divided into two n-1 The distance between the cameras in the nth layer is L0 / 2. Therefore, the distance between the cameras in the second layer of the pyramid, i.e., cameras 2, 6, 9, and 13, is L0 / 2, and cameras 2 and 6 are evenly distributed on the left and right sides of camera 4, and cameras 9 and 13 are evenly distributed on both sides of camera 11. The third layer is cameras 1, 3, 5, 7, 8, 10, 12, and 14, and the distance between the cameras is L0 / 4.

[0101] A three-dimensional imaging system based on a compound eye camera consisting of three layers of pyramids Figure 1 Take the example to explain the relationship between the camera field of view angles at different levels. Assume that the horizontal reference field of view angle of cameras No. 4 and No. 11 on the first level of the pyramid is selected as θ0 = 90 degrees, then the field of view angles of cameras No. 2, 6, 9, and 13 on the second level of the pyramid are To ensure that the field of view of all cameras in the second layer can cover the field of view of the cameras in the first layer of the pyramid. Similarly, the relationship between the field of view of the camera in the nth layer and the n-1th layer is:

[0102] A three-dimensional imaging system based on a compound eye camera consisting of three layers of pyramids Figure 5 Take 1 as an example to illustrate the relationship between camera sensor resolutions at different levels. In order to reduce the amount of calculations at higher levels of the pyramid, the resolution of the camera at the higher level can be appropriately reduced (using sensors with lower physical resolution or digital downsampling). The resolution relationship between the first and second layers of the pyramid is: R n ≥R n / 2. Figure 1 For example, due to Figure 1 The cameras in the middle and higher levels only supplement and correct the depth images in the horizontal direction of the lower level, so the second and third level cameras should have the same resolution as the first level camera in the vertical direction. In order to reduce the amount of calculation, the horizontal resolution of the higher level camera is greater than 1 / 2 of the corresponding lower level camera in the horizontal direction. For example, if the resolution of the first level camera is W*H, the resolution of the second level camera should be at least W / 2*H, and the resolution of the third level camera should be at least W / 4*H.

[0103] A three-dimensional imaging system based on a compound eye camera consisting of three layers of pyramids Figure 5 Take the frame rate relationship of camera sensors at different levels as an example to illustrate. When the level is higher, a higher frame rate can be used to detect motion. Suppose the frame rate of the pyramid level n is F n , then the frame rate of the n+1th layer is set to 2*F n .

[0104] It can be seen that in this optional embodiment, before acquiring the depth image to be processed, the position and number of cameras (target acquisition devices) deployed in each network level can be comprehensively considered based on the device type (main and secondary camera types) of the target acquisition device, combined with the number of target acquisition devices in each network level in the focal pyramid network and the acquisition direction of each network level, so that the position deployment of each camera meets the image acquisition requirements of the focal pyramid network. Such precise position deployment and hierarchical and sequential deployment improve the acquisition accuracy and reliability of the depth images acquired subsequently, thereby improving the processing efficiency and processing reliability of the subsequent processing of the acquired depth images, which is beneficial to improving the image quality of the target depth image finally processed.

[0105] Embodiment 2

[0106] See also Figure 2 , Figure 2 1 is a flow chart of another method for three-dimensional imaging of a compound eye camera based on a focal length pyramid network disclosed in an embodiment of the present invention. Figure 2 The described three-dimensional imaging method of a compound eye camera based on a focal length pyramid network can be applied to a three-dimensional imaging device of a compound eye camera based on a focal length pyramid network, and the embodiments of the present invention are not limited thereto. Figure 2 As shown, the compound eye camera three-dimensional imaging method based on the focal length pyramid network may include the following operations:

[0107] 201. Determine device parameters corresponding to a target acquisition device for acquiring depth images according to the determined focal length pyramid network.

[0108] 202. Determine the acquisition control parameters for all target acquisition devices based on the focal pyramid network.

[0109] 203. According to the device parameters and the acquisition control parameters, perform device parameter configuration operations and acquisition deployment operations on all target acquisition devices, and obtain the target depth images to be processed collected by all target acquisition devices.

[0110] In the embodiment of the present invention, for other descriptions of step 201 to step 203, please refer to other specific descriptions of step 101 to step 103 in embodiment 1, and the embodiment of the present invention will not be repeated here.

[0111] 204. Perform an initial depth calculation operation on all target depth images according to the focal length pyramid network to obtain initial depth calculation results corresponding to all target depth images.

[0112] 205. For each target depth image acquired by the target acquisition device, a motion information optimization operation is performed on the initial depth calculation result corresponding to the target depth image according to the determined motion information optimization algorithm to obtain a motion information optimized image corresponding to the initial depth calculation result corresponding to the target depth image.

[0113] 206. According to the progressive optimization algorithm corresponding to the focal length pyramid network, it is determined whether there is an image to be corrected in all the motion information optimized images.

[0114] In an embodiment of the present invention, the image to be corrected is an image corresponding to an area in a motion information optimized image that matches a preset correction area, wherein the preset correction area may be a binocular parallax occlusion area, a weak texture area, a repeated texture area, etc., which is not limited in the embodiment of the present invention.

[0115] In the embodiment of the present invention, when the judgment result of step 206 is yes, step 207 is executed, and when the judgment result of step 206 is no, step 208 is executed.

[0116] 207. According to the regional correction algorithm, and in accordance with the correction type corresponding to the image to be corrected, a deep correction optimization operation is performed on the image to be corrected, so that the image to be corrected is adjusted and optimized to meet a preset deep processing requirement.

[0117] 208. Determine that all motion optimized images are images that meet preset deep processing requirements, as deep processing results corresponding to the target depth images corresponding to the target acquisition device.

[0118] In the embodiment of the present invention, when the result of step 206 is , step 207 is executed, and the corresponding image processing logic is to adopt a bottom-up image optimization logic. In the depth image calculation of the first-level binocular stereo matching, there may be more error matching areas, such as binocular parallax occlusion areas, weak texture areas, repeated texture areas, etc. Therefore, in the calculation of each frame of the depth image, a higher level is used to optimize the error step by step. Taking the binocular parallax occlusion area as an example, since the base distance between cameras 4 and 11 at the first level is large, when the distance between the target and the camera is relatively close, when camera 4 is the main camera, a larger parallax occlusion area is generated on the left side of the main camera. However, at this time, cameras 1 and 3 at the third level are on the left side of the main camera 4, which partially solves the occlusion problem physically. The higher the pyramid level, the smaller the area of ​​the occlusion area. For the main camera of the first level, it is a repeated texture, but in the higher-level cameras, because more cameras with narrower fields of view are used, the repeated texture is decomposed, solving the repeated texture problem.

[0119] It can be seen that implementation Figure 2 The described three-dimensional imaging method of the compound eye camera based on the focal pyramid network can simultaneously process depth images of different network levels in a parallel computing manner, thereby improving the processing efficiency of depth image processing; it can also use motion information as an optimization parameter to perform motion information optimization on the initial depth calculation result, thereby improving the image quality and image accuracy of the target depth image finally processed; in addition, in the case where there is a correction area in the motion information optimization image, it can automatically perform a deep correction optimization operation on the image to be corrected according to the type to be corrected corresponding to the image to be corrected, thereby improving the image quality and image accuracy of the target depth image finally processed.

[0120] In an optional embodiment, for each target depth image acquired by the target acquisition device, a motion information optimization operation is performed on the initial depth calculation result corresponding to the target depth image according to the determined motion information optimization algorithm, and a method for obtaining a motion information optimized image corresponding to the initial depth calculation result corresponding to the target depth image specifically includes:

[0121] For each target depth image acquired by the target acquisition device, a motion information analysis operation is performed on the target depth image to obtain motion information between adjacent frame images in the target depth image, and a feature matching optimization operation is performed according to the motion information and the aligned motion information to obtain a feature matching optimization result corresponding to the motion information and the aligned motion information as the motion information optimized image corresponding to the target acquisition device;

[0122] The alignment motion information is information obtained after the alignment device performs a motion information analysis operation on the target depth image corresponding to the alignment device, and the alignment device is an alignment acquisition device corresponding to the acquisition device.

[0123] Optionally, the type to be corrected corresponding to the image to be corrected includes a binocular parallax occlusion type, a weak texture type or a repeated texture type;

[0124] Among them, in the focal length pyramid network, the image acquisition parameters corresponding to all target acquisition devices of each network level and all target acquisition devices of the network level adjacent to the network level meet the preset multiple conditions, wherein the multiple conditions are specifically:

[0125] The quotient of the resolution corresponding to the target acquisition device in the first level and the resolution corresponding to the target acquisition device in the second level is greater than a first preset value, wherein the network level corresponding to the first level is higher than the network level corresponding to the second level, and the first level is adjacent to the second level, and the first preset value is a positive number and less than 1;

[0126] The quotient of the image acquisition frame rate corresponding to the target acquisition device in the first level and the image acquisition frame rate corresponding to the target acquisition device in the second level is equal to a second preset value, and the second preset value is a positive number and greater than 1.

[0127] In this optional embodiment, for a specific description of the multiple condition, please refer to Example 1. Figure 5 The relevant description corresponding to the three-dimensional imaging system of the compound eye camera including the three-layer pyramid will not be repeated in this optional embodiment.

[0128] It can be seen that in this optional embodiment, the initial target depth result is optimized by using motion information. At the same time, as the network level increases, the resolution of the image sensor can be lower, and the motion of each camera frame information can be used to obtain more accurate feature points, which is used to improve the quality of the depth image obtained by using two cameras within each level.

[0129] Embodiment 3

[0130] See also Figure 3 , Figure 3It is a structural schematic diagram of a three-dimensional imaging device of a compound eye camera based on a focal pyramid network disclosed in an embodiment of the present invention. Among them, the three-dimensional imaging device of a compound eye camera based on a focal pyramid network can be a three-dimensional imaging terminal of a compound eye camera based on a focal pyramid network, a three-dimensional imaging device of a compound eye camera based on a focal pyramid network, a three-dimensional imaging system of a compound eye camera based on a focal pyramid network, or a three-dimensional imaging server of a compound eye camera based on a focal pyramid network. The three-dimensional imaging server of a compound eye camera based on a focal pyramid network can be a local server, a remote server, or a cloud server (also called a cloud server). When the three-dimensional imaging server of a compound eye camera based on a focal pyramid network is a non-cloud server, the non-cloud server can communicate with the cloud server, which is not limited in the embodiment of the present invention. Figure 3 As shown, the compound eye camera three-dimensional imaging device based on the focal length pyramid network may include a determination module 301, a deployment control module 302, an acquisition module 303 and a deep processing module 304, wherein:

[0131] Determination module 301 is used to determine the device parameters corresponding to the target acquisition device for acquiring depth images according to the determined focal length pyramid network, the device parameters including the number of devices and the image acquisition parameters corresponding to each target acquisition device, wherein the number of devices is an even number.

[0132] Optionally, the image acquisition parameters corresponding to each target acquisition device include a resolution and an image acquisition frame rate corresponding to the target acquisition device:

[0133] The focal pyramid network includes multiple network levels, and the collection control parameters for all target collection devices also include the number of network layers, wherein each network level is deployed with a number of target collection devices that matches the network level, the number of target collection devices between adjacent network levels is a preset multiple relationship, and the number of devices in the lower network level is less than the number of devices in the higher network level.

[0134] It should be noted that in the focal pyramid network, a preset parallel computing mode is used for target depth images at different network levels;

[0135] In the same network level of the focal pyramid network, each target acquisition device has at least one corresponding alignment acquisition device in the network level, and the target acquisition device and all the alignment acquisition devices corresponding to the target acquisition device use a parallel computing mode when processing the target depth images collected by each of them.

[0136] Determination module 301 is also used to determine the acquisition control parameters for all target acquisition devices based on the focal pyramid network. The acquisition control parameters include layout parameters and acquisition direction. The layout parameters are used to deploy the image acquisition position of each target acquisition device. The acquisition direction includes a vertical acquisition direction and / or a horizontal acquisition direction.

[0137] The deployment control module 302 is used to perform device parameter configuration operations and collection deployment operations on all target collection devices according to device parameters and collection control parameters.

[0138] The acquisition module 303 is used to acquire the target depth images to be processed acquired by all target acquisition devices, where all target depth images are images for the same acquisition target.

[0139] The deep processing module 304 is used to perform deep processing operations on all target depth images according to the focal length pyramid network to obtain deep processing results corresponding to all target depth images. The deep processing operations include motion information optimization operations.

[0140] It can be seen that implementation Figure 3 The described compound eye camera three-dimensional imaging device based on the focal length pyramid network can configure and deploy each acquisition device based on the pyramid network, and then use the pyramid network to perform deep processing on the depth image collected by each acquisition device, thereby realizing the correction, supplementation and optimization of the depth image, which is beneficial to reducing the proportion of blurred areas in the depth image, improving the image quality of the final depth image, and also improving the accuracy and reliability of the obtained depth image.

[0141] In an optional embodiment, the device type of the target acquisition device includes a main camera type and a secondary camera type, and the target acquisition devices corresponding to different device types perform different depth image acquisition operations. The deployment control module 302 performs device parameter configuration operations and acquisition deployment operations on all target acquisition devices according to device parameters and acquisition control parameters, specifically including:

[0142] According to the number of devices, determine the number of target acquisition devices that need to be deployed at each network level in the focal pyramid network;

[0143] According to the number of levels of each network level, determine the number of main cameras of the main camera type target acquisition devices and the number of secondary cameras of the secondary camera type target acquisition devices in the network level;

[0144] For each network level, the acquisition direction is used as the reference deployment direction. According to the number of main cameras and secondary cameras corresponding to the network level, the acquisition deployment operation is performed according to the image acquisition position of each target acquisition device in the network level, so that the deployment of each target acquisition device in the network level meets the deployment requirements corresponding to the network level.

[0145] After determining that all target collection devices in each network layer have completed the collection deployment operation, the device parameter configuration operation is performed on the target collection device according to the device parameters of each target collection device.

[0146] It can be seen that in this optional embodiment, before acquiring the depth image to be processed, the position and number of cameras (target acquisition devices) deployed in each network level can be comprehensively considered based on the device type (main and secondary camera types) of the target acquisition device, combined with the number of target acquisition devices in each network level in the focal pyramid network and the acquisition direction of each network level, so that the position deployment of each camera meets the image acquisition requirements of the focal pyramid network. Such precise position deployment and hierarchical and sequential deployment improve the acquisition accuracy and reliability of the depth images acquired subsequently, thereby improving the processing efficiency and processing reliability of the subsequent processing of the acquired depth images, which is beneficial to improving the image quality of the target depth image finally processed.

[0147] In another optional embodiment, the deep processing module 304 performs deep processing operations on all target depth images according to the focal pyramid network, and the manner in which the deep processing results corresponding to all target depth images are obtained specifically includes:

[0148] According to the focal length pyramid network, an initial depth calculation operation is performed on all target depth images to obtain initial depth calculation results corresponding to all target depth images;

[0149] For each target depth image acquired by the target acquisition device, a motion information optimization operation is performed on the initial depth calculation result corresponding to the target depth image according to the determined motion information optimization algorithm to obtain a motion information optimized image corresponding to the initial depth calculation result corresponding to the target depth image;

[0150] According to the progressive optimization algorithm corresponding to the focal length pyramid network, it is determined whether there is an image to be corrected in all the motion information optimized images, where the image to be corrected is an image corresponding to an area in a certain motion information optimized image that matches a preset correction area;

[0151] When it is determined that there is an image to be corrected among all the motion information optimized images, a deep correction optimization operation is performed on the image to be corrected according to the regional correction algorithm and the type to be corrected corresponding to the image to be corrected, so that the image to be corrected is adjusted and optimized to meet the preset deep processing requirements;

[0152] When it is determined that there is no image to be corrected in all motion information optimized images, all motion optimized images are determined to be images that meet the preset deep processing requirements as the deep processing results corresponding to the target depth images corresponding to the target acquisition device.

[0153] It can be seen that in this optional embodiment, depth images of different network levels can be processed simultaneously in a parallel computing manner, thereby improving the processing efficiency of processing depth images; it is also possible to use motion information as an optimization parameter to perform motion information optimization on the initial depth calculation result, thereby improving the image quality and image accuracy of the target depth image finally processed; in addition, in the case where there is a correction area in the motion information optimized image, it is possible to automatically perform a deep correction optimization operation on the image to be corrected according to the type to be corrected corresponding to the image to be corrected, thereby improving the image quality and image accuracy of the target depth image finally processed.

[0154] In another optional embodiment, the deep processing module 304 performs a motion information optimization operation on the initial depth calculation result corresponding to the target depth image collected by each target acquisition device according to the determined motion information optimization algorithm, and obtains a motion information optimized image corresponding to the initial depth calculation result corresponding to the target depth image. Specifically, the method includes:

[0155] For each target depth image acquired by the target acquisition device, a motion information analysis operation is performed on the target depth image to obtain motion information between adjacent frame images in the target depth image, and a feature matching optimization operation is performed according to the motion information and the aligned motion information to obtain a feature matching optimization result corresponding to the motion information and the aligned motion information as the motion information optimized image corresponding to the target acquisition device;

[0156] The alignment motion information is information obtained after the alignment device performs a motion information analysis operation on the target depth image corresponding to the alignment device, and the alignment device is an alignment acquisition device corresponding to the acquisition device.

[0157] It should be noted that the types of corrections corresponding to the images to be corrected include binocular parallax occlusion type, weak texture type or repeated texture type;

[0158] Among them, in the focal length pyramid network, the image acquisition parameters corresponding to all target acquisition devices of each network level and all target acquisition devices of the network level adjacent to the network level meet the preset multiple conditions, wherein the multiple conditions are specifically:

[0159] The quotient of the resolution corresponding to the target acquisition device in the first level and the resolution corresponding to the target acquisition device in the second level is greater than a first preset value, wherein the network level corresponding to the first level is higher than the network level corresponding to the second level, and the first level is adjacent to the second level, and the first preset value is a positive number and less than 1;

[0160] The quotient of the image acquisition frame rate corresponding to the target acquisition device in the first level and the image acquisition frame rate corresponding to the target acquisition device in the second level is equal to a second preset value, and the second preset value is a positive number and greater than 1.

[0161] It can be seen that in this optional embodiment, the initial target depth result can be optimized by using motion information. At the same time, as the network level increases, the resolution of the image sensor can be lower, and the motion of each camera frame information can be used to obtain more accurate feature points, which is used to improve the quality of the depth image obtained by using two cameras within each level.

[0162] Embodiment 4

[0163] See also Figure 5 , Figure 5 is a schematic diagram of the structure of another compound eye camera three-dimensional imaging device based on a focal length pyramid network disclosed in an embodiment of the present invention. Figure 5 As shown, the compound eye camera three-dimensional imaging device based on the focal length pyramid network may include:

[0164] A memory 401 storing executable program codes;

[0165] a processor 402 coupled to the memory 401;

[0166] The processor 402 calls the executable program code stored in the memory 401 to execute the steps in the three-dimensional imaging method of the compound eye camera based on the focal pyramid network described in the first embodiment of the present invention or the second embodiment of the present invention.

[0167] Embodiment 5

[0168] An embodiment of the present invention discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute the steps in the three-dimensional imaging method of a compound eye camera based on a focal pyramid network described in Embodiment 1 or Embodiment 2 of the present invention.

[0169] Embodiment 6

[0170] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps of the three-dimensional imaging method of a compound eye camera based on a focal pyramid network described in Example 1 or Example 2.

[0171] The device embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative work.

[0172] Through the specific description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution can be essentially or partly contributed to the prior art in the form of a software product, and the computer software product can be stored in a computer storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0173] Finally, it should be noted that the three-dimensional imaging method and device of a compound eye camera based on a focal pyramid network disclosed in the embodiment of the present invention only discloses a preferred embodiment of the present invention, which is only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A three-dimensional imaging method of a compound eye camera based on a focal pyramid network, characterized in that: The method comprises: Determine, according to the determined focal length pyramid network, device parameters corresponding to the target acquisition device for acquiring the depth image, wherein the device parameters include the number of devices and image acquisition parameters corresponding to each of the target acquisition devices, wherein the number of devices is an even number; Determine acquisition control parameters for all target acquisition devices according to the focal length pyramid network, wherein the acquisition control parameters include layout parameters and acquisition directions, wherein the layout parameters are used to deploy image acquisition positions of each target acquisition device, and the acquisition directions include vertical acquisition directions and / or horizontal acquisition directions; According to the device parameters and the acquisition control parameters, perform device parameter configuration operations and acquisition deployment operations on all the target acquisition devices, and obtain target depth images to be processed acquired by all the target acquisition devices, where all the target depth images are images for the same acquisition target; According to the focal length pyramid network, a deep processing operation is performed on all the target depth images to obtain deep processing results corresponding to all the target depth images, wherein the deep processing operation includes a motion information optimization operation; The image acquisition parameters corresponding to each of the target acquisition devices include the resolution and image acquisition frame rate corresponding to the target acquisition device: The focal length pyramid network includes multiple network levels, and the collection control parameters for all the target collection devices also include the number of network layers, wherein each network level is deployed with a number of target collection devices that matches the network level, the number of target collection devices between adjacent network levels is a preset multiple relationship, and the number of devices in the lower network level is less than the number of devices in the higher network level; In the focal pyramid network, a preset parallel computing mode is adopted for the target depth images at different network levels; In the same network level of the focal length pyramid network, each target acquisition device has at least one corresponding alignment acquisition device in the network level, and the target acquisition device and all the alignment acquisition devices corresponding to the target acquisition device use the parallel computing mode when processing the target depth images acquired by each of them; The performing a deep processing operation on all the target depth images according to the focal length pyramid network to obtain deep processing results corresponding to all the target depth images includes: According to the focal length pyramid network, an initial depth calculation operation is performed on all the target depth images to obtain initial depth calculation results corresponding to all the target depth images; For the target depth image acquired by each of the target acquisition devices, a motion information optimization operation is performed on the initial depth calculation result corresponding to the target depth image according to the determined motion information optimization algorithm to obtain a motion information optimized image corresponding to the initial depth calculation result corresponding to the target depth image; According to the progressive optimization algorithm corresponding to the focal length pyramid network, it is determined whether there is an image to be corrected in all the motion information optimized images, wherein the image to be corrected is an image corresponding to an area in a certain motion information optimized image that matches a preset correction area; When it is determined that the image to be corrected exists in all the motion information optimized images, a deep correction optimization operation is performed on the image to be corrected according to the area correction algorithm and the correction type corresponding to the image to be corrected, so that the image to be corrected is adjusted and optimized to meet the preset deep processing requirements; When it is determined that the image to be corrected does not exist in all the motion information optimized images, determining all the motion information optimized images as images that meet the preset deep processing requirements as the deep processing results corresponding to the target depth image corresponding to the target acquisition device; For each target depth image acquired by the target acquisition device, a motion information optimization operation is performed on the initial depth calculation result corresponding to the target depth image according to the determined motion information optimization algorithm to obtain a motion information optimized image corresponding to the initial depth calculation result corresponding to the target depth image, including: For the target depth image acquired by each of the target acquisition devices, a motion information analysis operation is performed on the target depth image to obtain motion information between adjacent frame images in the target depth image, and a feature matching optimization operation is performed according to the motion information and the alignment motion information to obtain a feature matching optimization result corresponding to the motion information and the alignment motion information as a motion information optimized image corresponding to the target acquisition device; The alignment motion information is information obtained after the alignment device performs the motion information analysis operation on the target depth image corresponding to the alignment device, and the alignment device is the alignment acquisition device corresponding to the acquisition device.

2. The three-dimensional imaging method of a compound eye camera based on a focal pyramid network according to claim 1, characterized in that: The device type of the target acquisition device includes a main camera type and a secondary camera type, and the target acquisition devices corresponding to different device types perform different depth image acquisition operations; The performing of device parameter configuration operations and collection deployment operations on all the target collection devices according to the device parameters and the collection control parameters includes: According to the number of devices, determining the number of levels of the target acquisition devices required to be deployed at each network level in the focal pyramid network; According to the number of levels of each of the network levels, determine the number of main cameras of the target acquisition devices of the main camera type and the number of secondary cameras of the target acquisition devices of the secondary camera type in the network level; For each of the network levels, taking the acquisition direction as the reference deployment direction, performing acquisition deployment operations according to the number of main cameras and the number of secondary cameras corresponding to the network level and the image acquisition position of each of the target acquisition devices in the network level, so that the deployment of each of the target acquisition devices in the network level meets the deployment requirements corresponding to the network level; After determining that all the target acquisition devices in each of the network levels have completed the acquisition deployment operation, a device parameter configuration operation is performed on the target acquisition device according to the device parameters of each target acquisition device.

3. The three-dimensional imaging method of compound eye camera based on focal pyramid network according to claim 1, characterized in that: The type of correction corresponding to the image to be corrected includes a binocular parallax occlusion type, a weak texture type or a repeated texture type; Among them, in the focal length pyramid network, the image acquisition parameters corresponding to all the target acquisition devices of each network level and all the target acquisition devices of the network level adjacent to the network level respectively meet the preset multiple conditions, wherein the multiple conditions are specifically: The quotient of the resolution corresponding to the target acquisition device in the first level and the resolution corresponding to the target acquisition device in the second level is greater than a first preset value, wherein the network level corresponding to the first level is higher than the network level corresponding to the second level, and the first level is adjacent to the second level, and the first preset value is a positive number and less than 1; The quotient of the image acquisition frame rate corresponding to the target acquisition device in the first level and the image acquisition frame rate corresponding to the target acquisition device in the second level is equal to a second preset value, and the second preset value is a positive number and greater than 1.

4. A three-dimensional imaging device of a compound eye camera based on a focal pyramid network, characterized in that: The device is used to execute the compound eye camera three-dimensional imaging method based on the focal pyramid network according to any one of claims 1 to 3, and the device comprises: A determination module, used to determine device parameters corresponding to a target acquisition device for acquiring a depth image according to the determined focal length pyramid network, wherein the device parameters include the number of devices and image acquisition parameters corresponding to each of the target acquisition devices, wherein the number of devices is an even number; The determination module is further used to determine the acquisition control parameters for all the target acquisition devices according to the focal length pyramid network, the acquisition control parameters including layout parameters and acquisition directions, the layout parameters are used to deploy the image acquisition position of each of the target acquisition devices, and the acquisition direction includes a vertical acquisition direction and / or a horizontal acquisition direction; A deployment control module, used to perform device parameter configuration operations and collection deployment operations on all the target collection devices according to the device parameters and the collection control parameters; An acquisition module, used to acquire target depth images to be processed acquired by all the target acquisition devices, wherein all the target depth images are images for the same acquisition target; A deep processing module is used to perform deep processing operations on all the target depth images according to the focal length pyramid network to obtain deep processing results corresponding to all the target depth images, wherein the deep processing operations include motion information optimization operations.

5. A three-dimensional imaging device of a compound eye camera based on a focal pyramid network, characterized in that: The device comprises: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the three-dimensional imaging method of the compound eye camera based on the focal pyramid network as described in any one of claims 1-3.

6. A computer storage medium, characterized in that: The computer storage medium stores computer instructions, which, when called, are used to execute the three-dimensional imaging method of the compound eye camera based on the focal pyramid network as described in any one of claims 1 to 3.

Images