CN110740238A - light splitting HDR camera applied to mobile robot SLAM field - Google Patents

Abstract

The invention discloses a spectroscopic HDR camera applied to the field of mobile robot SLAM, comprising: a casing and a lens mounted on the front end of the casing; a spectroscopic system is arranged in the casing, including: being arranged directly behind the lens The main beam splitter, the beam splitting surface of the main beam splitter is inclined to the optical axis of the lens, the rear of the main beam splitter is provided with a first secondary beam splitter, and the beam splitting surface of the first secondary beam splitter is perpendicular to the beam splitting surface of the main beam splitter; A second pair of beam splitters is arranged above the main beam splitter, and the beam splitting surface of the second pair beam splitter is parallel to the mirror surface of the main beam splitter; below and behind the first pair of beam splitters, and behind and above the second pair of beam splitters are respectively arranged A CMOS chip soldered on a PCB substrate, all PCB substrates are connected to an FPGA chip. The present invention can reduce the overall exposure time and reduce the occurrence of image smear phenomenon, and is suitable for large outdoor scenes and small indoor scenes due to the use of exposure methods in the same time and space.

Description

A Spectral HDR Camera Applied in the Field of Mobile Robot SLAM

technical field

The invention relates to the field of positioning and navigation of mobile robots, in particular to a high dynamic range imaging (HDR) camera applied to the field of simultaneous positioning and map construction (SLAM) of mobile robots.

Background technique

Image-based localization algorithm is a popular problem in the field of autonomous mobile robots, and it is a prerequisite for solving the motion planning and control of mobile robots. Image-based Simultaneous Localization and Mapping (SLAM) can recover a map of the environment while estimating the robot's own pose by analyzing image sequences in an unknown environment. Since this technology only relies on the image sensor of the robot itself, it does not require environmental modification and manual marking, and the camera has the advantages of low cost, so SLAM technology has received extensive attention at home and abroad. The main sensor of visual SLAM is the camera, so the image quality of the camera is directly related to the positioning accuracy of SLAM. However, due to the limited light latitude of traditional cameras, in some environments, such as sunlight through windows, shadows of objects, etc., the camera The obtained image will have partial overexposure and partial underexposure, losing most of the bright or dark information of the image, which is not conducive to the extraction of feature points by the SLAM algorithm, and the accuracy of feature extraction is directly related to the accuracy of the entire SLAM algorithm. . Although high-end cameras have high light latitude, their cost is often high, and it is very uneconomical to apply to mobile robots. Although ordinary cameras can obtain high dynamic range imaging (High Dynamic Range) through continuous exposure in a static environment. Imaging, HDR) images, however, due to the motion of the mobile robot, it will cause serious smearing of ordinary cameras in HDR mode. Therefore, it is of great significance to propose a cost-effective high dynamic range camera for the application of SLAM.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a spectroscopic HDR camera applied in the field of mobile robot SLAM, which is mainly used to solve the problem of motion blur in traditional cameras in HDR mode.

In order to realize the above-mentioned tasks, the present invention adopts the following technical solutions:

A spectroscopic HDR camera applied in the field of mobile robot SLAM, comprising: a casing and a lens mounted on the front end of the casing, a spectroscopic system is arranged in the casing, including:

The main beam splitter is arranged directly behind the lens, the beam splitter of the main beam splitter is inclined to the optical axis of the lens, a first sub beam splitter is arranged behind the main beam splitter, and the beam splitter of the first sub beam splitter is connected to the main beam The beam splitter surface of the mirror is vertical; a second pair of beam splitters is arranged above the main beam splitter, and the beam splitter surface of the second pair beam splitter is parallel to the mirror surface of the main beam splitter;

A first CMOS chip is arranged below the first sub-spectroscope, and a second CMOS chip is arranged behind the first sub-spectroscope; a third CMOS chip is arranged behind the second sub-spectroscope, and the second sub-spectroscope is arranged behind A fourth CMOS chip is arranged above the mirror; wherein, each CMOS chip is respectively welded on a PCB substrate; the PCB substrate corresponding to each CMOS chip is connected to the FPGA chip; wherein, the FPGA chip is used for each The images acquired by the CMOS chip are synthesized into high dynamic range images.

Further, the reflection and projection ratios of the main beam splitter, the first sub beam splitter, and the second sub beam splitter are all 1:1.

Further, the included angle between the beam splitting surface of the main beam splitter and the optical axis of the lens is 45°.

Further, the main beam splitter, the first sub beam splitter and the second sub beam splitter are all cube beam splitters.

After the outside light is focused by the lens and passes through the main beam splitter, 50% of the outside light penetrates the beam splitting surface of the main beam splitter and enters the first secondary beam splitter, and the remaining 50% of the outside light is reflected by the beam splitter surface of the main beam splitter. into the second pair of beamsplitters;

50% of the outside light entering the first pair of beamsplitters, half of the outside light is reflected by the beam splitting surface of the first pair of beamsplitters and then enters the first CMOS chip, and the other half of the outside light passes through the beam splitting of the first pair of beamsplitters The second CMOS chip is injected behind the face;

The remaining 50% of the external light entering the second sub-beamsplitter, half of which are reflected by the beam splitting surface of the second sub-beamsplitter and then enter the third CMOS chip, and the other half of the ambient light passes through the second sub-beamsplitter. After the beam splitting surface, the fourth CMOS chip is injected.

Further, the FPGA chip communicates with the outside through a data communication interface; when the FPGA chip receives an image request sent from the outside, it simultaneously issues exposure instructions with different exposure durations to the first CMOS chip to the fourth CMOS chip. To the fourth CMOS chips, images are acquired independently through the corresponding beam splitting surfaces.

Further, the working modes of the spectroscopic HDR camera include a low-light mode and a high-light mode, wherein the FPGA chip determines the average brightness and exposure time of the normally exposed image in the last image obtained by using each CMOS chip. The light conditions of the environment, and then choose to switch the working mode to low light mode or strong light mode before each CMOS chip acquires an image next time.

Further, the FPGA chip combines the images obtained by each CMOS chip into a high dynamic range image by means of coefficient fusion, including:

For an image acquired by each CMOS chip, each pixel of each image corresponds to a fusion coefficient, and then the high dynamic range image is obtained by weighted calculation of pixel values at the same position on different images.

Further, in the low light mode, the exposure instruction is: extend the exposure time by 2 steps, extend the exposure time by 1 step, normal exposure, and reduce the exposure time by 1 step, at this time, 2 overexposed images and 1 normal exposure image can be obtained. 1 image and 1 underexposed image, the spectroscopic system pays more attention to the image details in the dark part;

In the strong light mode, the exposure instructions are: extend the exposure time by 1 step, normal exposure, reduce the exposure time by 1 step, and reduce the exposure time by 2 steps. At this time, 1 overexposed image, 1 normal exposure image, If there are 2 underexposed images, the spectroscopic system pays more attention to the image details of the bright parts.

Further, a spectroscopic system bracket is arranged inside the housing, and the spectroscopic system bracket includes a pair of symmetrically arranged support plates, and the pair of support plates are symmetrically provided with a pair of supporting plates for fixing the main beam splitter and the first auxiliary beam splitter. L-shaped slot for mirror and second beam splitter;

After the main spectroscope, the first secondary spectroscope and the second secondary spectroscope are installed in the card slots, the support plates below and behind the first secondary spectroscope are symmetrically provided with a first fixing slot and a second fixing slot, respectively. It is used to fixedly install the PCB substrate corresponding to the first CMOS chip and the PCB substrate corresponding to the second CMOS chip; the support plate behind and above the second sub-spectroscope is symmetrically provided with a third fixing groove and a fourth fixing groove The grooves are respectively used to fix and install the PCB substrate corresponding to the third CMOS chip and the PCB substrate corresponding to the fourth CMOS chip.

Further, the PCB substrate and the end faces of the adjacent secondary beam splitters are parallel and have the same area.

The present invention has the following technical characteristics:

1. The camera of the present invention has an integrated spectroscopic system structure, and adopts the method of multi-CMOS simultaneous exposure to collect images under different exposure parameters to synthesize HDR images. Compared with the traditional HDR imaging method, the overall exposure time can be reduced, and the image drag can be reduced. Due to the exposure method of the same time and space, it is suitable for large outdoor scenes and small indoor scenes.

2. The present invention uses multiple low-cost CMOS chips to obtain HDR images through synthesis, which has a huge cost advantage compared to chips with high light latitude, and helps reduce the cost of a mobile robot positioning system.

3. Compared with ordinary cameras, the camera of the present invention has a higher dynamic range, which helps the SLAM algorithm of the mobile robot to improve the accuracy of positioning and map construction.

Description of drawings

1 is a schematic structural diagram of the present invention after removing part of the shell;

Fig. 2 is the schematic diagram of the spectroscopic system support of the present invention;

Fig. 3 is the overall structure schematic diagram of the present invention;

Description of the symbols in the figure: 1 lens, 2 shell, 3 fourth CMOS chip, 4 second secondary beam splitter, 5FPGA chip, 6 circuit board, 7 beam splitting system bracket, 71 support plate, 72 card slot, 73 first fixing slot , 74 second fixing slot, 75 third fixing slot, 76 fourth fixing slot, 77 fifth fixing slot, 8 third CMOS chip, 9 second CMOS chip, 10 first secondary beam splitter, 11 first CMOS chip, 12 main beamsplitters.

Detailed ways

The present invention discloses a spectroscopic HDR camera applied to the SLAM neighborhood of a mobile robot. As shown in FIG. 1 , it includes: a casing 2 and a lens 1 installed at the front end of the casing 2 , for example, the lens 1 can be fixed by screw fitting At the front end of the housing 2; a spectroscopic system is arranged in the housing 2, including:

The main beam splitter 12 is arranged directly behind the lens 1, wherein the right rear means that the center line of the main beam splitter 12 coincides with the optical axis of the lens 1; the beam splitter surface of the main beam splitter 12 is inclined to the optical axis of the lens 1 , a first sub-spectroscope 10 is arranged behind the main beam splitter 12, and the beam splitting surface of the first sub beam splitter 10 is perpendicular to the beam splitter surface of the main beam splitter 12; a second sub beam splitter 4 is arranged above the main beam splitter 12, The beam splitting surface of the secondary beam splitter 4 is parallel to the mirror surface of the main beam splitter 12 . In this embodiment, the main beam splitter 12 , the first sub beam splitter 10 and the second sub beam splitter 4 are all square cube beam splitters with the same size, and the beam splitting surfaces are all on one bottom edge of the square cube and the other Between the side and top edges, the angle between the beam splitting surface and the top and bottom surfaces of the cube is 45°; the reflection and projection ratios of the main beam splitter 12, the first secondary beam splitter 10, and the second secondary beam splitter 4 are all 45°. 1:1, that is, after the light is irradiated on the beam splitter surface of the beam splitter in a direction parallel to the bottom surface of the beam splitter, half of the light is reflected by the beam splitter surface, while the other half of the light passes through the beam splitter surface.

Optionally, the included angle between the beam splitting surface of the main beam splitter 12 and the optical axis of the lens 1 is 45°, so as to accurately perform beam splitting. As shown in FIG. 1 , the lower end of the main beam splitter 12 is closer to the side of the lens 1 .

A first CMOS chip 11 is arranged below the first sub-beamsplitter 10 , a second CMOS chip 9 is arranged behind the first sub-beamsplitter 10 , and a third CMOS chip is arranged behind the second sub-beamsplitter 4 8. A fourth CMOS chip 3 is arranged above the second secondary beam splitter 4; wherein, each CMOS chip (including the first CMOS chip 11 to the fourth CMOS chip 3) is welded on a PCB substrate for receiving The light passing through the corresponding secondary beam splitter; the PCB substrate corresponding to each CMOS chip is connected to the FPGA chip 5; the corresponding PCB substrate refers to the PCB substrate to which the CMOS chip is welded, and the CMOS chip is realized on the PCB substrate. Peripheral circuits. The PCB substrate and the end faces of the adjacent sub-beamsplitters are parallel and have the same area. In this embodiment, each PCB substrate is a square plate, the size of which is the same as that of the first sub-beamsplitter 10 and the second sub-beamsplitter 4 . The end faces have the same area and the centers are on the same axis. For example, the PCB substrate corresponding to the first CMOS chip 11 is arranged in parallel directly below the first secondary beam splitter 10, and its size is the same as that of the bottom surface of the first secondary beam splitter 10, and the centers of the two are on the same axis; The PCB substrates corresponding to the two CMOS chips 9 are arranged in parallel directly behind the first sub-beam splitter 10 , and have the same size as the rear surface of the first sub-beam splitter 10 .

In this solution, after the outside light is focused by the lens 1 and passes through the main beam splitter 12, 50% of the outside light penetrates the beam splitting surface of the main beam splitter 12 and enters the first sub beam splitter 10, and the remaining 50% of the outside light The light is reflected by the beam splitting surface of the main beam splitter 12 into the second sub beam splitter 4; 50% of the external light entering the first sub beam splitter 10, half of which is reflected by the beam splitting surface of the first sub beam splitter 10 After entering the first CMOS chip 11, the other half of the external light passes through the beam splitting surface of the first secondary beam splitter 10 and then enters the second CMOS chip 9; the remaining 50% of the external light entering the second secondary beam splitter 4, wherein Half of the ambient light is reflected by the beam splitting surface of the second sub-beamsplitter 4 and then enters the third CMOS chip 8 , and the other half of the ambient light passes through the beam-splitting surface of the second sub beamsplitter 4 and then enters the fourth CMOS chip 3 . Therefore, after the external light passes through the light splitting system, there can be 25% of the external light on each CMOS chip.

In this solution, the FPGA chip 5 is fixed inside the housing 2 through the circuit board 6, and the FPGA chip 5 communicates with the outside through a data communication interface; the PCB substrate is connected to the circuit board through a cable, and then communicates with the FPGA chip 5; The interface can be, for example, USB3.0, and the external can be, for example, a computer, a controller of a mobile robot, and the like. The FPGA chip 5 is used for synthesizing the under-exposure, normal-exposure, and over-exposure images acquired by each CMOS chip into a high-dynamic-range image.

When the FPGA chip 5 receives the image request sent from the outside, it simultaneously issues exposure commands with different exposure durations to the first CMOS chip 11 to the fourth CMOS chip 3 , and the first CMOS chip 11 to the fourth CMOS chip 3 pass through the corresponding beam splitters respectively. The first CMOS chip 11 and the second CMOS chip 9 obtain images independently through reflection and transmission of the beam splitting surface of the first secondary beam splitter 10, respectively, while the third CMOS chip 8 and the fourth CMOS chip 3 pass through the The reflection and transmission of the beam splitting surface of the second sub beam splitter 4 independently acquire images.

The working modes of the spectroscopic HDR camera include low-light mode and high-light mode, wherein the FPGA chip 5 determines the environmental conditions by analyzing the average brightness of the normally exposed image and the exposure time in the last image obtained by using each CMOS chip. light conditions, and then choose to switch the working mode to low light mode or strong light mode before each CMOS chip acquires an image next time. The image acquired by each CMOS chip is processed and synthesized to finally obtain the high dynamic range image.

In low light mode, the exposure instructions are: extend the exposure time by 2 steps, extend the exposure time by 1 step, normal exposure, and reduce the exposure time by 1 step. At this time, 2 overexposed images, 1 normal exposure image, One underexposed image, in this case, the spectroscopic system pays more attention to the image details in the dark part;

In the strong light mode, the exposure instructions are: extend the exposure time by 1 step, normal exposure, reduce the exposure time by 1 step, and reduce the exposure time by 2 steps. At this time, 1 overexposed image, 1 normal exposure image, There are 2 underexposed images, in which case the spectroscopic system pays more attention to the image details of the bright parts.

Among them, the 1st file and the 2nd file refer to the degree of exposure time. For example, extending the exposure time by 1 file means extending the exposure time T, then extending the exposure time by 2 files is extending the exposure time 2T; the same is true for reducing the exposure time; Gears can be set according to actual needs.

The acquisition of the normal exposure time is based on the principle of maximum entropy of the image. The image entropy reflects the average amount of information in the image. After each acquisition of a normal exposure image, the FPGA chip 5 counts the current image entropy value, and then according to the entropy value to adjust the next exposure time. Image entropy is calculated using the following formula:

where p _i is the probability that a pixel with pixel gray value i appears in the image:

p _i =n _i /M×N

Among them, M and N represent the number of rows and columns of pixels in the image, respectively, and n _i represents the number of pixels with a grayscale value of i in the image.

The FPGA chip 5 combines the images obtained by each CMOS chip into a high dynamic range image by means of coefficient fusion. Specifically, the fusion weights of different exposure images are estimated first, and then weighted average fusion is performed. The fusion method follows the following formula:

where W ⁿ (x, y) is the fusion weight, I ⁿ (x, y) represents different images, and n represents the image sequence number.

That is, for an image obtained by each CMOS chip, each pixel of each image corresponds to a fusion coefficient, and then the high dynamic range image is obtained by weighted calculation of pixel values at the same position on different images.

As shown in FIG. 1 , in order to facilitate the arrangement of the spectroscopic system, a spectroscopic system bracket 7 is provided inside the housing 2 , and the spectroscopic system bracket 7 includes a pair of support plates 71 arranged symmetrically. The symmetrical arrangement refers to this A pair of support plates 71 are parallel to each other and correspond to each other; the pair of support plates 71 are symmetrically provided with L-shaped slots for fixing the main beam splitter 12, the first sub beam splitter 10 and the second sub beam splitter 4; for example As shown in FIG. 2 , since the main beam splitter 12 , the first sub beam splitter 10 , and the second sub beam splitter 4 are positioned behind the main beam splitter 12 , the second sub beam splitter 4 is located in the main beam splitter 4 Above the spectroscope 12, the L-shaped slot formed on the support plate 71 can well fix these spectroscopes; into the L-shaped slot on the pair of support plates 71 .

After the main beam splitter 12 , the first sub beam splitter 10 and the second sub beam splitter 4 are installed in the card slots, the support plates 71 below and behind the first sub beam splitter 10 are symmetrically provided with first fixing grooves 73 , The second fixing slot 74, as shown in FIG. 1 and FIG. 2, both the first fixing slot 73 and the second fixing slot 74 are strip-shaped slots, which are respectively used for fixing and installing the PCB substrate corresponding to the first CMOS chip and the second fixing slot 74. The PCB substrate corresponding to the CMOS chip; specifically, the two sides of the PCB substrate are respectively snapped into the fixing grooves on the pair of support plates 71, that is, the PCB substrate is effectively fixed, and the CMOS on the PCB substrate is effectively fixed. The chip is effectively fixed so that it can receive the light passing through the beamsplitter at the precise position set. A third fixing slot 75 and a fourth fixing slot 76 are symmetrically opened on the support plate 71 behind and above the second secondary beam splitter 4 , respectively for fixing the PCB substrate corresponding to the third CMOS chip and the fourth fixing slot 76 . The PCB substrate corresponding to the CMOS chip.

Further, as shown in FIG. 2 , the support plate 71 on the right side of the fourth fixing slot 76 is also symmetrically provided with a fifth fixing slot 77 . Referring to FIG. 1 , the fifth fixing slot is used for the circuit board 6 of the FPGA chip 5 .

In this embodiment, when the lens 1 and the housing 2 are threadedly connected, the thread type is the common C-type interface of the industrial lens 1, and the parameters of the lens 1 are: focal length 5mm, aperture 1.8, and the maximum image surface suitable for CMOS 1/ 2", horizontal viewing angle 64°, vertical viewing angle 50°, lens 1 distortion -0.33%, lens 1 length 58mm, lens 1 maximum resolution 10M, lens 1 interface C-mount.

The first CMOS chip 11 to the fourth CMOS chip 3 are of the same model, which is ON Semiconductor MT9V034C12STM in this embodiment, with an image resolution of 752H×480V, a maximum image surface of 1/3”, a pixel size of 6.0um×6.0um, and grayscale imaging. , global shutter, maximum frame rate 60Hz, 10-bit ADC resolution, dynamic range greater than 55dB. The cost of this CMOS is lower, and at the same time, it has a higher pixel size, which helps to improve the dynamic range of the image. Each CMOS chip corresponds to The specifications of the PCB substrate are the same, for example, the thickness of the PCB substrate is 1.2mm and the size is 20mm×20mm.

The main beam splitter 12 , the first beam splitter, and the second beam splitter have the same specifications, and the preferred size is 20 mm×20 mm×20 mm.

The FPGA chip 5 connects and communicates with the outside through the USB3.0 interface, so as to send exposure instructions to each CMOS chip, receive, process, and synthesize images at the same time, and send the final synthesized image to the outside through the USB3.0 interface. For example, the model of the FPGA chip 5 may be: Xilinx Zynq 7020, wherein the programmable logic unit is 85K, and it has sufficient computing resources to process data of four CMOS chips. At the same time, the chip contains a dual-core ARM Cortex-A9 SOC computing unit, which can perform general calculations to calculate image entropy, exposure time, etc. The FPGA chip 5 measures the current environment after each acquisition of the image data of the CMOS chip, counts the grayscale distribution of the normal exposure image, that is, the probability of each grayscale value from 0 to 255, and then calculates the information entropy of the normal exposure image. , follow the principle of maximum information entropy, and dynamically adjust the normal exposure time of the next frame. According to the normal exposure time of the next frame, it is determined that the current mode is in the low light mode and the strong light mode.

This embodiment uses 4 low-level CMOS chips with high cost performance and a light splitting system to achieve simultaneous exposure to the environment at the same position, and HDR images are obtained through image synthesis, which solves the problem of insufficient light latitude of images captured by traditional cameras and problems of HDR by ordinary cameras. The motion blur problem in the mode is suitable for small indoor scenes and large outdoor scenes. The cost advantage of higher-end cameras is prominent, which helps to improve the SLAM accuracy of mobile robots.

Claims (10)

1. A spectroscopic HDR camera applied to the field of mobile robot SLAM, comprising: a casing (2) and a lens (1) mounted on the front end of the casing (2), wherein a spectroscopic system is provided in the casing (2) , characterized in that the spectroscopic system includes: The main beam splitter (12) is arranged directly behind the lens (1), the beam splitting surface of the main beam splitter (12) is inclined to the optical axis of the lens (1), and a first beam splitter (12) is arranged behind the main beam splitter (12). A secondary beam splitter (10), the light splitting surface of the first secondary beam splitter (10) is perpendicular to the beam splitting surface of the main beam splitter (12); a second secondary beam splitter (4) is arranged above the main beam splitter (12), The beam splitting surfaces of the second pair of beam splitters (4) are parallel to the mirror surface of the main beam splitter (12); A first CMOS chip (11) is provided below the first secondary beam splitter (10), and a second CMOS chip (9) is provided behind the first secondary beam splitter (10); the second secondary beam splitter (10) is provided with a second CMOS chip (9). 4) a third CMOS chip (8) is arranged at the rear, and a fourth CMOS chip (3) is arranged above the second secondary beam splitter (4); wherein, each CMOS chip is respectively welded on a PCB substrate; each The PCB substrate corresponding to one CMOS chip is connected to the FPGA chip (5); wherein, the FPGA chip (5) is used for synthesizing the image acquired by each CMOS chip into a high dynamic range image. 2. The spectroscopic HDR camera applied to the field of mobile robot SLAM according to claim 1, wherein the main beam splitter (12), the first secondary beam splitter (10), the second secondary beam splitter (4) The ratio of reflection and projection is 1:1. 3. The spectroscopic HDR camera applied in the field of mobile robot SLAM according to claim 1, wherein the main beam splitter (12), the first secondary beam splitter (10) and the second secondary beam splitter (4) All are cube beamsplitters. 4. The spectroscopic HDR camera applied to the field of mobile robot SLAM as claimed in claim 1, wherein the angle between the spectroscopic surface of the main beam splitter (12) and the optical axis of the lens (1) is 45 °; After the outside light is focused by the lens (1) and passes through the main beam splitter (12), 50% of the outside light penetrates the beam splitting surface of the main beam splitter (12) and enters the first sub beam splitter (10), and the remaining 50% % of the external light is reflected by the beam splitting surface of the main beam splitter (12) into the second sub beam splitter (4); 50% of the ambient light entering the first secondary beam splitter (10), half of the ambient light is reflected by the beam splitting surface of the first secondary beam splitter (10) and then injected into the first CMOS chip (11), and the other half of the ambient light enters the first CMOS chip (11) After passing through the beam splitting surface of the first secondary beam splitter (10), it is injected into the second CMOS chip (9); The remaining 50% of the ambient light entering the second sub-beamsplitter (4), half of the ambient light is reflected by the beam splitting surface of the second sub-beamsplitter (4) and then enters the third CMOS chip (8), and the other half of the ambient light enters the third CMOS chip (8) The light enters the fourth CMOS chip (3) after passing through the beam splitting surface of the second secondary beam splitting mirror (4). 5. The spectroscopic HDR camera applied to the field of mobile robot SLAM as claimed in claim 1, wherein the FPGA chip (5) communicates with the outside through a data communication interface; when the FPGA chip (5) receives an externally sent When an image is requested, exposure commands with different exposure durations are issued to the first CMOS chip (11) to the fourth CMOS chip (3) at the same time, and the first CMOS chip (11) to the fourth CMOS chip (3) respectively pass through the corresponding light splitting surfaces Acquire images independently. 6. The spectroscopic HDR camera applied to the field of mobile robot SLAM as claimed in claim 1, wherein the working mode of the spectroscopic HDR camera comprises a weak light mode and a strong light mode, wherein the FPGA chip (5) is analyzed by analyzing In the image obtained by each CMOS chip last time, the average brightness of the normally exposed image and the exposure time are used to determine the light condition of the environment, and then before each CMOS chip captures the image next time, choose to switch the working mode to low light mode or Strong light mode. 7. The spectroscopic HDR camera applied to the field of mobile robot SLAM as claimed in claim 5, wherein in the low light mode, the exposure instruction is: extend the exposure time by 2 steps, extend the exposure time by 1 step, normal Expose and reduce the exposure time by 1 stop, at this time, 2 overexposed images, 1 normal exposure image, and 1 underexposed image can be obtained, and the spectroscopic system pays more attention to the image details in the dark part; In the strong light mode, the exposure instructions are: extend the exposure time by 1 step, normal exposure, reduce the exposure time by 1 step, and reduce the exposure time by 2 steps. At this time, 1 overexposed image, 1 normal exposure image, If there are 2 underexposed images, the spectroscopic system pays more attention to the image details of the bright parts. 8. The spectroscopic HDR camera applied in the field of mobile robot SLAM according to claim 1, wherein the FPGA chip (5) synthesizes the image obtained by each CMOS chip into a high dynamic range image by means of coefficient fusion ,include: For an image acquired by each CMOS chip, each pixel of each image corresponds to a fusion coefficient, and then the high dynamic range image is obtained by weighted calculation of pixel values at the same position on different images. 9. The spectroscopic HDR camera applied in the field of mobile robot SLAM according to claim 1, wherein the shell (2) is provided with a spectroscopic system bracket (7), and the spectroscopic system bracket (7) comprises A pair of support plates (71) are symmetrically arranged, and the pair of support plates (71) are symmetrically provided with a pair of support plates (71) for fixing the main beam splitter (12), the first sub beam splitter (10) and the second sub beam splitter ( 4) L-shaped card slot; After the main beam splitter (12), the first sub beam splitter (10) and the second sub beam splitter (4) are installed in the card slot, the support plates (71) below and behind the first sub beam splitter (10) A first fixing groove (73) and a second fixing groove (74) are symmetrically opened on the upper part, which are respectively used for fixing and installing the PCB substrate corresponding to the first CMOS chip (11) and the PCB substrate corresponding to the second CMOS chip (9). ; A third fixing slot (75) and a fourth fixing slot (76) are symmetrically provided on the rear and upper support plates (71) of the second sub-spectroscope (4), respectively for fixing and installing the third fixing slot (76). A PCB substrate corresponding to the CMOS chip (8) and a PCB substrate corresponding to the fourth CMOS chip (3). 10 . The spectroscopic HDR camera applied in the field of mobile robot SLAM according to claim 1 , wherein the PCB substrate and the end faces of the adjacent secondary beam splitters are parallel and have the same area. 11 .

Images