A hexagonal grid-based intelligent system for mass customization of camera arrays

Abstract

The present invention discloses a large-scale customized camera array intelligent system based on a hexagonal grid, comprising a server, four clients and N customized cameras arranged in a hexagonal grid; the four clients are connected to the server in communication based on the TCP / IP network protocol; when N is an integer multiple of 4, each client is connected to N / 4 customized cameras via a HUB through an independent USB3.2 bus, and the N / 4 customized cameras include a root camera and (N / 4‑1) sub-cameras. The client is used to perform contour extraction on the image data from the customized camera to obtain the data to be reproduced and send it to the server, and the server executes a hardware-assisted image-based visual shell algorithm after receiving the data to be reproduced to realize the reproduction of a three-dimensional dynamic scene. The present invention can reduce the ghosting of the output image and realize the real-time presentation of the three-dimensional dynamic scene.

Description

[Technical field]

[0001] The invention relates to the field of virtual reality, and in particular to a large-scale customized camera array intelligent system based on a hexagonal grid. [Background technology]

[0002] In the field of virtual reality, key technologies for stereoscopic display have made rapid progress. The generation of stereoscopic display effects requires binocular parallax, motion parallax and adaptive adjustment of focus. In order to create binocular parallax and motion parallax, two stereoscopic scene views corresponding to the left and right eyes are required; by providing multiple views for each eye, adaptive adjustment of focus can be further created. Therefore, it is very important to establish a method that can synthesize multiple arbitrary stereoscopic space views with high quality. Modern computer graphics technology can synthesize any number of three-dimensional geometric model views. For the human visual system, three-dimensional geometric models are not only convenient for stereoscopic viewing, but also conducive to physiological fusion. In the field of virtual reality, the virtual world is composed of geometric models of scene objects. If the same strategy is to be applied to the actual physical world, geometric models of real objects must be constructed. However, obtaining an accurate geometric model requires measuring the three-dimensional shape, surface texture and optical coefficients of the object, which takes too much manpower and is not so easy to obtain. In order to circumvent this problem, a technology called IBR (Image-Based Rendering) has emerged in recent years and has received a lot of attention.

[0003] IBR technology requires the integration of multiple images, which may come from the same or multiple different viewpoints, and may also need to be collected simultaneously or at different times. There are usually two ways to achieve such image acquisition, one is to use a single moving camera, and the other is to use an array of multiple cameras. For static scenes, both methods are feasible, but the first method takes a long time to collect; for dynamic scene acquisition, a camera array must be used as the input system.

[0004] In the design and application of camera arrays, it is recorded that Wang Kedong et al. (2005) designed a 4×4 CMOS (Complementary Metal-Oxide-Semiconductor) image sensor array to capture stroboscopic spark images with extremely high temporal resolution. C. Zhang et al. (2007) developed a self-reconfigurable camera array using 48 commercial Ethernet cameras with electronic control panels and horizontal movement, with the goal of improving the view interpolation effect by changing the position and direction of the camera according to the scene geometry and the desired virtual viewpoint. Matusik et al. (2000) used IBVH (Image-Based Visual Hull) technology to build a 4-camera array that supports reproduction. Yang et al. (2002) built a 5-camera system that supports real-time reproduction with the help of modern graphics hardware; Schirmacher et al. (2001) constructed a 6-camera system for real-time processing of general brightness graphics; Naemura et al. (2002) constructed a 16-camera system for real-time reproduction. Some large arrays consisting of dozens of cameras have also been built, such as Stanford's multi-camera array (2002), MIT's distributed light field camera array (2002), and CMU's 3D space (2002). These three systems consist of 128, 64, and 49 cameras respectively.

[0005] The development of camera arrays has created new opportunities for high-fidelity reproduction of dynamic scenes, but it has also brought challenges. In the cases of camera arrays mentioned above, most of them are for static scenes or objects. Those arrays composed of a small number of cameras are usually easy to obtain real-time reproduction, but the reproduction quality is poor. Although large camera arrays increase the viewpoint range, they generate a huge amount of data during the acquisition process. It is difficult to achieve satisfactory real-time presentation effects due to the need to process a large amount of data. For example, the Stanford system based on IEEE1394 (a serial standard developed by Apple) established in 2002 consists of 128 video cameras, but it is difficult to cope with the huge data throughput, so only a partial implementation of 6 cameras is actually given. In order to provide a more universal production platform for different applications, the camera array must be able to collect and process the huge amount of data generated in real time. The arrangement spacing between cameras is an important reason for the ghosting artifacts of the output image. Therefore, its design must fully consider the requirements of various applications for data transmission bandwidth and image quality. In addition, the unprocessed raw video acquisition data often exceeds the maximum sustained throughput of the computer's general hard disk and peripheral interface. In order to continuously transmit and store the video stream, the raw data must be compressed, and the compression ratio and image loss must be considered during compression.

[0006] In recent years, the rapid development of graphics hardware has made it possible to render visual shells directly from a set of contour images in real time. However, there are some drawbacks in existing visual shell calculations, such as the number of input images is limited by the graphics hardware specifications, reconstruction and reproduction are performed in steps, the algorithm is too cumbersome and the computation is too large. [Summary of the invention]

[0007] The technical problem to be solved by the present invention is to overcome the defects of the camera array in the prior art in that a huge amount of data is generated during the acquisition process, resulting in great difficulty in real-time processing, output image ghosting, and great pressure in transmitting and storing video streams. A scalable, hexagonal grid-based large-scale customized camera array high-bandwidth intelligent system is provided based on the USB3.2 (Universal Serial Bus 3.2) standard and C / S (client / server) architecture, and an improved graphics hardware-assisted image-based visual shell reproduction algorithm is used to synthesize high-fidelity high-frame rate output views, so as to fully realize the real-time acquisition and presentation of three-dimensional dynamic scenes.

[0008] The present invention solves the above technical problems through the following technical solutions:

[0009] The present invention provides a large-scale customized camera array intelligent system based on a hexagonal grid, which is characterized in that it includes a server, four clients and N customized cameras, N≥36 and N≤508, and N is a multiple of 6;

[0010] N of the custom cameras are arranged in a hexagonal grid, with each custom camera located at the center of each grid pixel;

[0011] The four client machines are connected to the server through communication based on TCP / IP (Transmission Control Protocol / Internet Protocol) network protocol;

[0012] When N is an integer multiple of 4, each of the client computers is connected to N / 4 of the customized cameras via an independent USB3.2 bus via a HUB (Universal Serial Bus Hub), and the N / 4 customized cameras include a root camera and (N / 4-1) sub-cameras;

[0013] When N is a non-integer multiple of 4, two of the clients are connected to (N+2) / 4 of the customized cameras through an independent USB3.2 bus via a HUB, and the (N+2) / 4 customized cameras include 1 root camera and ((N+2) / 4-1) sub-cameras; another two of the clients are connected to (N-2) / 4 of the customized cameras through a USB3.2 bus via a HUB, and the (N-2) / 4 customized cameras include 1 root camera and ((N-2) / 4-1) sub-cameras;

[0014] The root camera is used to generate a clock synchronization signal and a trigger signal and transmit them to the corresponding sub-camera via a USB3.2 bus;

[0015] The client is used to extract contours from image data from N customized cameras to obtain data to be reproduced and send the data to be reproduced to the server. After receiving the data to be reproduced, the server is used to execute a hardware-assisted image-based visual shell algorithm to achieve three-dimensional dynamic scene reproduction.

[0016] Preferably, the customized camera includes a first circuit board, a second circuit board, a CMOS image sensor, a DSP (digital signal processor), a CPLD (Complex Programmable Logic Device), a USB interface chip and a memory;

[0017] The first circuit board and the second circuit board are connected via a cable;

[0018] The CMOS image sensor, the CPLD, the USB interface chip and the memory are electrically connected to the DSP respectively; the CPLD is also electrically connected to the CMOS image sensor and the USB interface chip;

[0019] Each of the customized cameras is located at the center of each grid pixel, including: the first circuit board of each of the customized cameras is located at the center of each grid pixel; the CMOS image sensor is arranged on the first circuit board;

[0020] The DSP, the CPLD, the USB interface chip and the memory are arranged on the second circuit board;

[0021] The CMOS image sensor is used to collect images of dynamic scenes and transmit them to the DSP;

[0022] The DSP is used to process and compress the received image of the dynamic scene to obtain the image data;

[0023] The CPLD is used to control the timing of the customized cameras and output the clock synchronization signal to achieve synchronization between the customized cameras;

[0024] The USB interface chip is connected to the corresponding USB3.2 bus.

[0025] Preferably, the server is also used to issue at least one of the following commands: downloading the executable code and encoding of the DSP, establishing timing, programming the registers of the image sensor and recording the data to be reproduced and running the rendering program; the client is also used to run the array software to complete the control of the large-scale customized camera array and the storage of data.

[0026] Preferably, N is 36, and the 36 customized cameras are evenly disposed on six parallel linear guide rails located in the same plane and arranged in a hexagonal grid.

[0027] Preferably, the 36 customized cameras are used to synchronously capture images of dynamic scenes at a capture speed of 20 to 40 frames per second, and the resolution of the images of the dynamic scenes is greater than 320×240; when reproducing the three-dimensional dynamic scene, the resolution of the reproduced view is greater than 640×480, and the reproduction frame rate is 30 to 80 frames per second.

[0028] Preferably, the length of the customized camera is less than or equal to 4 centimeters, the width of the customized camera is less than or equal to 4 centimeters, the height of the customized camera is 8 to 18 centimeters, and the distance between the customized cameras is 4 to 8 centimeters.

[0029] Preferably, the length of the customized camera is 4 cm, the width of the customized camera is 4 cm, the height of the customized camera is 10 cm, and the distance between the customized cameras is 6 cm.

[0030] Preferably, the large-scale customized camera array intelligent system also includes a plurality of disk servers, each of the customized cameras is attached with a disk server, and the rotation range of the disk server is less than or equal to 90 degrees.

[0031] Preferably, the visual shell algorithm adopts a multi-process method of texture objects plus template buffer to process any number of input images; the visual shell algorithm also adopts a method based on color plus depth image to hide the reconstruction in the reproduction process.

[0032] Preferably, the compression algorithm adopted by the DSP includes at least one of MPEG-4 (a standard compression algorithm), M-JPEG (a standard compression algorithm), H.264 (a standard compression algorithm), and X264 (a standard compression algorithm).

[0033] The positive and progressive effects of the present invention are: the large-scale customized camera array intelligent system based on hexagonal grid provided by the present invention realizes a scalable customized camera array high-bandwidth intelligent system based on USB3.2 standard and mature C / S architecture, which can be flexibly expanded from 36 cameras to 508 cameras. The hexagonal grid structure adopted can eliminate the problem of output image ghosting, and the image-based visual shell reproduction algorithm assisted by improved graphics hardware can synthesize high-fidelity and high-frame-rate output views, so as to fully realize the real-time acquisition and presentation of three-dimensional dynamic scenes. The present invention realizes a customized camera array that can be automatically reconfigured and fully controlled by the program, so as to realize the effect of active arrangement and active acquisition of the array intelligent system according to the user's selection. The present invention can collect and process the huge amount of data generated by the array in real time, and has fully considered the requirements of various applications for data transmission bandwidth and image quality.

Brief Description of the Drawings

[0034] Figure 1 The figure is a schematic structural diagram of a hexagonal grid-based large-scale customized camera array intelligent system according to a preferred embodiment of the present invention.

[0035] Figure 2 FIG. 4 is a schematic diagram of a hexagonal grid camera array in a preferred embodiment of the present invention.

[0036] Figure 3 It is a functional block diagram of a customized camera in a preferred embodiment of the present invention.

[0037] Figure 4Another schematic diagram of a hexagonal grid camera array in a preferred embodiment of the present invention. [Specific implementation method]

[0038] The present invention is further described below by way of examples, but the present invention is not limited to the scope of the examples.

[0039] like Figure 1 As shown, this embodiment discloses a large-scale customized camera array intelligent system based on a hexagonal grid, including a server, four clients and N customized cameras 4, N ≥ 36 and N ≤ 508, N is a multiple of 6. The N customized cameras 4 are arranged as shown in Figure 2 The camera array 5 in a hexagonal grid is shown, and each custom camera 4 is located at the center of each grid pixel. Four clients are connected to the server based on the TCP / IP network protocol 2. When N is an integer multiple of 4, each client is connected to N / 4 custom cameras 4 via an independent USB3.2 bus 1 via a HUB (not shown in the figure), and the N / 4 custom cameras 4 include a root camera and (N / 4-1) sub-cameras.

[0040] When N is a non-integer multiple of 4, two of the four clients are connected to (N+2) / 4 custom cameras 4 through an independent USB3.2 bus 1 via a HUB, and the (N+2) / 4 custom cameras 4 include a root camera and ((N+2) / 4-1) sub-cameras; the other two of the four clients are connected to (N-2) / 4 custom cameras 4 through a USB3.2 bus 1 via a HUB, and the (N-2) / 4 custom cameras 4 include a root camera and ((N-2) / 4-1) sub-cameras. The root camera is used to generate a clock synchronization signal and a trigger signal and transmit them to the sub-cameras in the same USB3.2 bus 1 via the USB3.2 bus 1.

[0041] The client in this embodiment is used to extract the contours of the image data from the N customized cameras 4 to obtain the data to be reproduced and send the data to be reproduced to the server. After receiving the data to be reproduced, the server is used to execute the hardware-assisted image-based visual shell algorithm to achieve three-dimensional dynamic scene reproduction. Among them, the visual shell algorithm uses a multi-process method of texture objects plus template buffers to process any number of input images; the visual shell algorithm also uses a method based on color plus depth images to hide the reconstruction in the reproduction process.

[0042] When N is 42, 42 custom cameras 4 are evenly arranged on six parallel linear guides located in the same plane. Figure 4In this embodiment, N is 36, and 36 custom cameras 4 are evenly arranged on six parallel linear guide rails located in the same plane and arranged as follows: Figure 2 The hexagonal grid camera array 5 shown. 36 custom cameras are used to synchronously capture images of dynamic scenes, with a capture speed of 20 to 40 frames per second, and the resolution of the images of dynamic scenes is greater than 320×240; the view resolution of the three-dimensional dynamic scene reproduction is greater than 640×480, and the reproduction frame rate is 30 to 80 frames per second. In order to more effectively obtain dynamic scenes and dynamic objects under different circumstances, the camera array in this embodiment is a reconfigurable array, and the custom camera 4 on the guide rail can be moved horizontally, thereby realizing a custom camera array that can be automatically reconfigured and fully controlled by the program, thereby obtaining the effect of active arrangement and active acquisition of the large-scale custom camera array intelligent system based on the hexagonal grid according to the user's selection.

[0043] like Figure 3 As shown, the customized camera 4 in this embodiment includes a first circuit board 6, a second circuit board 7, a CMOS image sensor, a DSP, a CPLD, a USB interface chip and a memory; wherein the memory includes SDRAM (synchronous dynamic random-access memory) and FLASH (flash memory). The first circuit board 6 and the second circuit board 7 are connected by a cable.

[0044] The CMOS image sensor, CPLD, USB interface chip and memory are electrically connected to the DSP respectively; the CPLD is also electrically connected to the CMOS image sensor and the USB interface chip. The USB interface chip is connected to the corresponding USB3.2 bus. Each customized camera 4 is located at the center of each grid pixel, including: the first circuit board 6 of each customized camera 4 is located at the center of each grid pixel; the CMOS image sensor is arranged on the first circuit board 6. The remaining devices including the DSP, CPLD, USB interface chip and memory are arranged on the second circuit board 7.

[0045] In this embodiment, in order to ensure that the customized cameras are small enough to facilitate close arrangement between the cameras, the CMOS image sensor is customized as a small circuit board, while the image processing, transmission and other circuits are customized as a larger circuit board. The two circuit boards are connected by cables for signal connection.

[0046] The CMOS image sensor is used to collect images of dynamic scenes and transmit them to the DSP; the DSP is used to process and compress the received images of dynamic scenes to obtain image data; the CPLD is used to control the internal timing of the custom camera 4 and output a clock synchronization signal to achieve synchronization between N custom cameras 4. The compression algorithm used by the DSP includes at least one of MPEG-4, M-JPEG, H.264, and X264.

[0047] In this embodiment, the timing and synchronization control in the camera array is very critical and determines whether the video captured by the customized camera array can meet the needs of the application. Figure 2 As shown, the CPLD device in the custom camera provides accurate timing and synchronization between custom camera triggers. The server sends the establishment timing command through the client, and the unified clock and trigger are first sent to the root camera of each bus. The root camera buffers the clock and sends it to the next one, and so on. Each custom camera uses the system clock through a PLL (Phase Locked Loop) to generate a corrected working clock. Since the clock is not used for data transmission between boards (that is, the clock is independent of the transmission of data between boards), the camera-to-camera delay is not important, and the shared clock ensures that all boards are frequency locked. Having a common system clock means that all sensors and encoders are running at the same frequency. The cable that transmits the clock transmits the global trigger signal to all custom cameras on each bus. These signals are directly connected to the CPLD device on the custom camera, which controls the initial synchronization or interval sensor shutter timing and the precise start of all video streams or snapshot frames.

[0048] The length of the customized camera 4 is less than or equal to 4 cm, the width of the customized camera 4 is less than or equal to 4 cm, the height of the customized camera 4 is 8 to 18 cm, and the spacing between the customized cameras 4 is 4 to 8 cm. In this embodiment, the size of the customized camera is 4×4×10 cm, that is, 4 cm in length, 4 cm in width, and 10 cm in height. Since the customized camera array spacing is calculated from the center origin of the grid pixel, if the array spacing is 6 cm, the gap between the customized cameras is only 2 cm, so the specific range of the customized camera array spacing can be set to 4 to 8 cm, but it must be greater than 4 cm, otherwise there will be no gap between the customized cameras.

[0049] The large-scale customized camera array intelligent system disclosed in this embodiment also includes a plurality of disk servers. Each customized camera is attached with a standard panoramic disk server. The rotation scale of the disk server is less than or equal to 90 degrees and can rotate within a range of 90 degrees.

[0050] In this embodiment, the USB interface chip of each custom camera uploads the video data processed by the DSP to the respective client, and the client sends the instructions to the DSP through the USB3.2 bus interface, and the DSP controls the operation of the custom camera according to these instructions. The specific working process is: when the custom camera is connected to the array, an enumeration process is generated between the client and the custom camera, that is, the client detects that the custom camera is connected, automatically sends a query request, and the custom camera responds to this request and sends the device's Vendor ID (manufacturer identification) and Product ID (product identification). The client loads the corresponding device driver according to these two IDs (identifications) to complete the enumeration process. Then, the client sends instructions through the USB3.2 bus interface to configure and start the DSP. At this time, the image data input from the CMOS image sensor is processed and compressed by the DSP and sent to the client through the USB3.2 bus interface. The client performs operations such as contour extraction and distance conversion on the input image, and sends the processing results to the server for reproduction. For example, 36 video streams are synchronously captured at a resolution of 320×240 pixels, with an input frame rate of 30fps (frames per second), and output at a resolution of 640×480 pixels, with a reproduction frame rate of 30 to 80fps, reaching the level of real-time.

[0051] The large-scale customized camera array intelligent system based on hexagonal grid provided in this embodiment adopts the USB3.2 communication interface that is available in ordinary computers and is superior to IEEE1394 in many aspects, which reduces the development of interface boards and conforms to the development direction of computer high-speed interfaces; adopts a client / server architecture based on TCP / IP protocol to accommodate a larger number of cameras, providing an effective diversion mechanism for large-scale video transmission; adopts a hexagonal grid structure to reduce the problem of output image ghosting; adopts a customized camera including a CMOS image sensor to control the image resolution, frame rate and synchronization timing, etc., and can perform 90-degree panoramic rotation; and adopts visual shell calculation assisted by graphics hardware, which is conducive to fully tapping the increasingly powerful programmable performance of modern computer graphics hardware to achieve real-time frame rate and high-fidelity reproduction.

[0052] The large-scale customized camera array intelligent system based on hexagonal grid provided in this embodiment realizes a scalable customized camera array high-bandwidth intelligent system based on USB3.2 standard and mature C / S architecture, which can be flexibly expanded from 36 cameras to 508 cameras, and thus has considerable scalability. The hexagonal grid structure adopted can reduce the problem of output image ghosting, and the image-based visual shell reproduction algorithm assisted by improved graphics hardware can synthesize high-fidelity and high-frame-rate output views to fully realize the real-time acquisition and presentation of three-dimensional dynamic scenes. The present invention realizes a customized camera array that can be automatically reconfigured and fully controlled by the program to obtain the effect that the array intelligent system actively arranges and actively acquires according to the user's selection. The present invention can collect and process a huge amount of data generated by the array in real time, and has fully considered the requirements of various applications for data transmission bandwidth and image quality, so it can realize high-fidelity real-time reproduction of dynamic scenes, and can provide a universal production platform for practical applications such as three-dimensional television, interactive games, remote conferences, and large-screen stereo advertisements.

[0053] Although the specific embodiments of the present invention are described above, it should be understood by those skilled in the art that these are only examples, and the protection scope of the present invention is defined by the appended claims. Those skilled in the art may make various changes or modifications to these embodiments without departing from the principles and essence of the present invention, but these changes and modifications all fall within the protection scope of the present invention.

Claims

1. A large-scale customized camera array intelligent system based on hexagonal grid, characterized in that: It includes one server, four clients and N customized cameras, where N≥36 and N≤508, and N is a multiple of 6; N of the custom cameras are arranged in a hexagonal grid, with each custom camera located at the center of each grid pixel; The four clients are connected to the server through TCP / IP network protocol; When N is an integer multiple of 4, each of the client computers is connected to N / 4 customized cameras via a HUB through an independent USB3.2 bus, and the N / 4 customized cameras include a root camera and (N / 4-1) sub-cameras; When N is a non-integer multiple of 4, two of the clients are connected to (N+2) / 4 of the customized cameras through an independent USB3.2 bus via a HUB, and the (N+2) / 4 customized cameras include 1 root camera and ((N+2) / 4-1) sub-cameras; another two of the clients are connected to (N-2) / 4 of the customized cameras through a USB3.2 bus via a HUB, and the (N-2) / 4 customized cameras include 1 root camera and ((N-2) / 4-1) sub-cameras; The root camera is used to generate a clock synchronization signal and a trigger signal and transmit them to the corresponding sub-camera via a USB3.2 bus; The client is used to extract contours from the image data from the N customized cameras to obtain data to be reproduced and send the data to be reproduced to the server. After receiving the data to be reproduced, the server is used to execute a hardware-assisted image-based visual shell algorithm to achieve three-dimensional dynamic scene reproduction; The visual shell algorithm adopts a multi-process method of texture objects plus template buffer to process any number of input images; the visual shell algorithm also adopts a method based on color plus depth image to hide the reconstruction in the reproduction process.

2. The hexagonal grid-based large-scale customized camera array intelligent system according to claim 1, characterized in that: The customized camera includes a first circuit board, a second circuit board, a CMOS image sensor, a DSP, a CPLD, a USB interface chip and a memory; The first circuit board and the second circuit board are connected via a cable; The CMOS image sensor, the CPLD, the USB interface chip and the memory are electrically connected to the DSP respectively; The CPLD is also electrically connected to the CMOS image sensor and the USB interface chip; Each of the customized cameras is located at the center of each grid pixel, including: the first circuit board of each of the customized cameras is located at the center of each grid pixel; the CMOS image sensor is arranged on the first circuit board; The DSP, the CPLD, the USB interface chip and the memory are arranged on the second circuit board; The CMOS image sensor is used to collect images of dynamic scenes and transmit them to the DSP; The DSP is used to process and compress the received image of the dynamic scene to obtain the image data; The CPLD is used to control the timing of the customized cameras and output the clock synchronization signal to achieve synchronization between the customized cameras; The USB interface chip is connected to the corresponding USB3.2 bus.

3. The hexagonal grid-based large-scale customized camera array intelligent system as claimed in claim 2, characterized in that: The server is also used to issue at least one of the following commands: downloading the executable code and encoding of the DSP, establishing timing, programming the registers of the image sensor and recording the data to be reproduced, and running the rendering program; the client is also used to run the array software to complete the control of the large-scale customized camera array and the storage of data.

4. The hexagonal grid-based large-scale customized camera array intelligent system as claimed in claim 1, characterized in that: N is 36, and the 36 customized cameras are evenly arranged on six parallel linear guide rails located in the same plane and arranged in a hexagonal grid.

5. The hexagonal grid-based large-scale customized camera array intelligent system as claimed in claim 4, characterized in that: The 36 customized cameras are used to synchronously capture images of dynamic scenes at a capture speed of 20 to 40 frames per second. The resolution of the images of the dynamic scenes is greater than 320×240. The view resolution during reproduction of the three-dimensional dynamic scene is greater than 640×480, and the reproduction frame rate is 30 to 80 frames per second.

6. The hexagonal grid-based large-scale customized camera array intelligent system as claimed in claim 1, characterized in that: The length of the customized camera is less than or equal to 4 centimeters, the width of the customized camera is less than or equal to 4 centimeters, the height of the customized camera is 8 to 18 centimeters, and the distance between the customized cameras is 4 to 8 centimeters.

7. The hexagonal grid-based large-scale customized camera array intelligent system according to claim 6, characterized in that: The length of the customized camera is 4 cm, the width of the customized camera is 4 cm, the height of the customized camera is 10 cm, and the distance between the customized cameras is 6 cm.

8. The hexagonal grid-based large-scale customized camera array intelligent system as claimed in claim 1, characterized in that: The large-scale customized camera array intelligent system also includes a plurality of disk servers, each of the customized cameras is attached with a disk server, and the rotation range of the disk server is less than or equal to 90 degrees.

9. The hexagonal grid-based large-scale customized camera array intelligent system as claimed in claim 2, characterized in that: The compression algorithm adopted by the DSP includes at least one of MPEG-4, M-JPEG, H.264, and X264.

Images