Binocular imaging method and device based on low-illumination scene, terminal and storage medium

Abstract

The invention discloses a binocular imaging method and device based on a low-illumination scene, a terminal and a storage medium, and the method comprises the steps: collecting an initial scene image corresponding to an initial scene, and determining an illumination parameter according to the initial scene image; performing light supplement on the initial scene according to the illumination parameters, and determining a target scene; and collecting a target scene image corresponding to the target scene based on a binocular camera, and determining a depth map of the target scene according to the target scene image. The brightness degree of the initial scene is judged according to the initial scene image, so that the illumination parameter is obtained, and the area with insufficient illumination in the initial scene is supplemented with light according to the illumination parameter, so that a binocular camera can collect a high-quality image based on the scene after light supplement, and the problem that the image quality is poor due to insufficient illumination in the prior art can be effectively solved. And the surface texture features of the images captured by the left and right cameras are not obvious, so that the accuracy of depth calculation is reduced.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a binocular imaging method, device, terminal and storage medium based on low-light scenes. Background Art

[0002] In binocular stereo matching, algorithms usually rely on feature points provided by texture information to achieve matching. Lighting has an important impact on the texture information of the captured image. When the light is insufficient, the texture in the scene may be weakened or disappear, making the texture features of the object surface in the images captured by the left and right cameras unclear, thereby reducing the accuracy of depth calculation.

[0003] Therefore, the existing technology still needs to be improved and developed. Summary of the invention

[0004] The technical problem to be solved by the present invention is that, in view of the above-mentioned defects of the prior art, a binocular imaging method, device, terminal and storage medium based on low-light scenes are provided, aiming to solve the problem in the prior art that due to insufficient light, the surface texture features of the images captured by the left and right cameras are not obvious, thereby reducing the accuracy of depth calculation.

[0005] The technical solution adopted by the present invention to solve the problem is as follows: In a first aspect, an embodiment of the present invention provides a binocular imaging method based on a low-light scene, wherein the method comprises: Collecting an initial scene image corresponding to the initial scene, and determining illumination parameters according to the initial scene image; Filling the initial scene with light according to the illumination parameters to determine a target scene; A target scene image corresponding to the target scene is collected based on a binocular camera, and a depth map of the target scene is determined according to the target scene image.

[0006] In one implementation method, determining the illumination parameters according to the initial scene image includes: Determining a brightness distribution matrix according to the initial scene image, wherein the brightness distribution matrix is ​​used to represent the brightness of each position in the initial scene image; Determine the illumination intensity and illumination angle according to the brightness distribution matrix; The illumination parameter is determined according to the illumination intensity and the illumination angle.

[0007] In one implementation method, determining a brightness distribution matrix according to the initial scene image includes: Determine the position information and light intensity information corresponding to each pixel point according to the initial scene image; The brightness distribution matrix is ​​determined according to each of the position information and each of the light intensity information.

[0008] In one implementation method, determining the illumination intensity and illumination angle according to the brightness distribution matrix includes: Obtaining a preset brightness value, and determining a point to be filled in light corresponding to the initial scene according to the preset brightness value and the brightness distribution matrix; Acquire a preset range, and determine the light intensity according to the preset range and the point to be filled light; The illumination angle is determined according to a center point corresponding to the initial scene image and the light point to be filled.

[0009] In an implementation method, before collecting the target scene image corresponding to the target scene based on the binocular camera, it also includes: forming a random dot matrix on the target scene through a diffractive optical element.

[0010] In one implementation method, determining the depth map of the target scene according to the target scene image includes: The matching cost between the pixels corresponding to the target scene images respectively acquired by the left and right cameras of the binocular camera is calculated based on the similarity metric; Determine the aggregate matching cost corresponding to each pixel according to the aggregate window and each of the matching costs; Performing disparity calculation based on the aggregate matching cost to determine a disparity map; Obtain binocular camera parameters, and determine the depth map according to the binocular camera parameters and the disparity map.

[0011] In one implementation method, before determining the depth map according to the binocular camera parameters and the disparity map, the method includes: The disparity map is subjected to disparity post-processing, wherein the disparity post-processing method includes: one or more of left-right consistency detection, disparity filling, and median filtering.

[0012] In a second aspect, an embodiment of the present invention further provides a binocular imaging device based on a low-light scene, wherein the binocular imaging device based on a low-light scene comprises: A parameter determination module, used to collect an initial scene image corresponding to the initial scene, and determine the illumination parameters according to the initial scene image; A brightness adjustment module, used to fill in the initial scene with light according to the illumination parameters to determine the target scene; The binocular imaging module is used to collect a target scene image corresponding to the target scene based on a binocular camera, and determine a depth map of the target scene according to the target scene image.

[0013] In a third aspect, an embodiment of the present invention further provides a terminal, comprising a memory and one or more processors; the memory stores one or more programs; the program comprises instructions for executing any of the binocular imaging methods based on low-light scenes as described above; and the processor is used to execute the program.

[0014] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium on which a plurality of instructions are stored, wherein the instructions are suitable for being loaded and executed by a processor to implement any of the above-mentioned binocular imaging methods based on low-light scenes.

[0015] Beneficial effects of the present invention: The embodiments of the present invention collect an initial scene image corresponding to the initial scene, determine illumination parameters based on the initial scene image; fill in the illumination for the initial scene based on the illumination parameters to determine the target scene; collect a target scene image corresponding to the target scene based on a binocular camera, and determine a depth map of the target scene based on the target scene image. Since the present invention determines the brightness of the initial scene based on the initial scene image, thereby obtaining illumination parameters, and fills in the insufficiently illuminated areas in the initial scene based on the illumination parameters, so that the binocular camera can collect high-quality images based on the scene after fill in, it can effectively solve the problem in the prior art that the surface texture features of the images captured by the left and right cameras are not obvious due to insufficient illumination, thereby reducing the accuracy of depth calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 It is a flow chart of a binocular imaging method based on a low-light scene provided by an embodiment of the present invention.

[0018] Figure 2 It is a schematic diagram of the internal modules of a binocular imaging device based on a low-light scene provided by an embodiment of the present invention.

[0019] Figure 3 It is a principle block diagram of a terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The present invention discloses a binocular imaging method, device, terminal and storage medium based on low-light scenes. In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0021] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0022] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as generally understood by those skilled in the art in the art to which the present invention belongs. It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with the meanings in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless specifically defined as herein.

[0023] In binocular stereo matching, algorithms usually rely on feature points provided by texture information to achieve matching. Lighting has an important impact on the texture information of the captured image. When the light is insufficient, the texture in the scene may be weakened or disappear, making the texture features of the object surface in the images captured by the left and right cameras unclear, thereby reducing the accuracy of depth calculation.

[0024] In view of the above-mentioned defects of the prior art, the present invention provides a binocular imaging method based on low-light scenes, wherein the method collects an initial scene image corresponding to the initial scene, determines the illumination parameters according to the initial scene image; fills the initial scene with illumination according to the illumination parameters to determine the target scene; collects the target scene image corresponding to the target scene based on the binocular camera, and determines the depth map of the target scene according to the target scene image. Since the present invention determines the brightness of the initial scene according to the initial scene image, thereby obtaining the illumination parameters, and fills the insufficiently illuminated area in the initial scene according to the illumination parameters, so that the binocular camera can collect high-quality images based on the scene after fill-in illumination, it can effectively solve the problem in the prior art that the surface texture features of the images captured by the left and right cameras are not obvious due to insufficient illumination, thereby reducing the accuracy of the depth calculation.

[0025] Exemplary Methods The binocular imaging method based on low-light scenes of this embodiment can be used on a device with a binocular module. The binocular module in this embodiment includes a binocular camera, a DOE (diffractive optical element) laser module, an image sensor, and a central processing unit.

[0026] like Figure 1 As shown, the method includes: Step S100: Acquire an initial scene image corresponding to an initial scene, and determine illumination parameters according to the initial scene image.

[0027] Lighting is an important factor in binocular stereo imaging and has a significant impact on depth calculation. Insufficient lighting may cause the texture in the scene to weaken or disappear, making the texture features of the image surface captured by the left and right cameras of the binocular camera unclear, resulting in inaccurate depth calculation. In addition, in scenes with insufficient lighting, there may also be problems with uneven lighting, resulting in parallax distortion in the images captured by the left and right cameras, and ultimately leading to incorrect estimated depth, causing the position of objects in the depth map to shift. Therefore, in this embodiment, an initial scene image corresponding to the initial scene is collected, and the initial scene image is analyzed to determine which positions in the current scene need additional lighting, etc., so as to obtain lighting parameters, wherein the lighting parameters include one or more of light intensity and lighting angle.

[0028] In one implementation, the light parameters may be determined based on the perceived brightness of the environment sensed by a photosensitive sensor.

[0029] In one implementation, determining the illumination parameters according to the initial scene image includes: Step S101, determining a brightness distribution matrix according to the initial scene image, wherein the brightness distribution matrix is ​​used to represent the brightness of each position in the initial scene image; Step S102, determining the illumination intensity and illumination angle according to the brightness distribution matrix; Step S103: determining the illumination parameter according to the illumination intensity and the illumination angle.

[0030] This embodiment obtains a brightness distribution matrix based on the brightness information of each position in the collected initial image. In one implementation, the brightness distribution matrix is ​​determined based on the initial scene image, including: step S1011, determining the position information and light intensity information corresponding to each pixel point respectively based on the initial scene image; step S1012, determining the brightness distribution matrix based on each position information and each light intensity information. In this embodiment, an image sensor is used to read the position information and light intensity information of each pixel point in the collected initial scene image when collecting the initial scene image, and after normalizing the light intensity information of each pixel point, a brightness distribution matrix is ​​generated based on each position information and each light intensity information. Among them, in addition to outputting position information and light intensity information through image sensors, there are many other methods to determine the position information and light intensity information corresponding to each pixel point according to the initial scene image, such as converting the initial scene image into a grayscale image, and determining the position information and light intensity information of each pixel point according to the coordinate value and grayscale value of each pixel point in the grayscale image, or the initial scene image is a color image, and determining the brightness and contrast of the pixel point according to the pixel grayscale values ​​of the color image in the three channels of R channel, G channel, and B channel, thereby determining the light intensity information corresponding to the pixel point according to the brightness and contrast, etc.

[0031] This embodiment analyzes the brightness distribution matrix and the brightness of each position in the initial scene to obtain the insufficiently illuminated positions in the initial scene and the required supplementary lighting intensity of the initial scene, thereby obtaining the lighting angle and lighting intensity, and determines the lighting parameters based on the lighting angle and lighting intensity, increases lighting for the areas with lower brightness in the initial scene, improves the texture information of subsequently collected images, and enhances the accuracy of pixel matching of images.

[0032] In one implementation, the brightness distribution matrix can be determined by a histogram method in image processing. For example, a number of local image blocks corresponding to the initial scene image are determined based on the initial scene image, a histogram (color histogram or grayscale histogram) corresponding to each local image block is calculated, light intensity information of each local image block is determined according to each histogram, and a brightness distribution matrix is ​​determined according to position information of each local image block in the initial scene image and corresponding light intensity information.

[0033] In one implementation, determining the illumination intensity and illumination angle according to the brightness distribution matrix includes: Step S1021, obtaining a preset brightness value, and determining a point to be filled light corresponding to the initial scene according to the preset brightness value and the brightness distribution matrix; Step S1022, obtaining a preset range, and determining the light intensity according to the preset range and the point to be filled light; Step S1023: determining the illumination angle according to the center point corresponding to the initial scene image and the light point to be filled.

[0034] Specifically, this embodiment sets different preset brightness values ​​according to different methods for generating the brightness distribution matrix (different ways of obtaining light intensity information and position information). The difference between each brightness value in the brightness distribution matrix and the preset brightness value is calculated to obtain a brightness difference matrix, wherein the difference between the brightness value in the brightness distribution matrix greater than the preset brightness value is a positive number, and the difference between the brightness value in the brightness distribution matrix less than the preset brightness value is a negative number. By finding the minimum value in the brightness difference matrix, it can be known that the pixel corresponding to the minimum value is the position with the lowest brightness in the initial scene image, so the pixel position corresponding to the minimum value is used as the light point to be filled. With the light point to be filled as the center, the difference within the preset range is obtained, the average value of the difference within the preset range is calculated, and the illumination intensity is determined according to the difference of the light point to be filled and the average value of the difference within the preset range. In one implementation, the deep learning method in the central processing unit is used in advance to obtain a high-quality texture image as the goal, and the mapping relationship between the difference of the light to be filled, the average value of the difference within the preset range and the illumination intensity of the light to be filled is learned, so as to determine the illumination intensity of the light to be filled according to the learned mapping relationship.

[0035] Specifically, the fill light instrument and the image sensor are both installed in the middle of the binocular camera. Since the low light in the scene may be an area, it is considered that a light beam can be irradiated into the low light area to supplement the light for the low light area. The method of determining the lighting angle according to the center point corresponding to the initial scene image and the light point to be filled in this embodiment specifically includes: obtaining the center point coordinates according to the center point corresponding to the initial scene image, constructing a two-dimensional coordinate system according to the center point coordinates, taking the center point coordinates as the origin of the two-dimensional coordinate system, and the x-axis and y-axis of the two-dimensional coordinate system are parallel to the length and width of the image respectively. According to the fill light point coordinates of the light point to be filled in the initial scene image, the quadrant in which the light point to be filled in the two-dimensional coordinate system formed by the x-axis and the y-axis is located is determined, and the lighting angle is determined according to the quadrant in which it is located. For example, when the light point to be filled is in the first quadrant, the illumination angle is set to 45° in the upper right; when the light point to be filled is in the second quadrant, the illumination angle is set to 45° in the upper left; when the light point to be filled is in the third quadrant, the illumination angle is set to 45° in the lower left; when the light point to be filled is in the fourth quadrant, the illumination angle is set to 45° in the lower right. This embodiment sets the illumination angle with fill light by judging the quadrant where the light point to be filled is located. The calculation is simple and fast, and does not require a large amount of computing resources. At the same time, it can also supplement the illumination for the position with insufficient illumination intensity in the scene, thereby improving the texture information of the subsequently collected image.

[0036] like Figure 1 As shown, the method further comprises the following steps: Step S200: fill in the initial scene with light according to the illumination parameters to determine a target scene.

[0037] Simply put, lighting parameters include lighting intensity and lighting angle. The lighting intensity and lighting angle of the fill-light instrument are set according to the lighting parameters to fill-light the low-light areas in the initial scene so as to obtain higher quality original images in subsequent acquisitions.

[0038] In one implementation, a DOE laser module is installed in the middle of the binocular camera. The diffraction infrared fill light technology of the DOE laser module is used to increase the illumination of the environment. In dark scenes or when sunlight is not enough to cover up the infrared light, the texture created by the infrared irradiation of the DOE speckle laser on the object is used to achieve a binocular stereo matching effect that takes into account both indoor and outdoor scenes. This opens up new possibilities for high-precision three-dimensional reconstruction under complex lighting and texture conditions.

[0039] In one implementation, before collecting the target scene image corresponding to the target scene based on the binocular camera, the method further includes: forming a random dot matrix on the target scene through a diffractive optical element.

[0040] Since fill light usually comes from a single direction, it may cause problems such as uneven brightness and overexposure. To further improve the matching accuracy, this embodiment introduces infrared random speckles from DOE speckle lasers to create additional textures for each object in the scene, helping the binocular camera to capture more feature points in an environment with insufficient texture. The texture complexity of the image is increased, thereby enhancing the reliability of the binocular matching algorithm, especially the semi-global matching algorithm.

[0041] like Figure 1 As shown, the method further comprises the following steps: Step S300: collecting a target scene image corresponding to the target scene based on a binocular camera, and determining a depth map of the target scene according to the target scene image.

[0042] A depth map, also called a distance image, is an image that uses the distance value (depth value) from the image collector to each point in the scene as a pixel value. In this embodiment, the left and right cameras of a binocular camera are used to capture the target scene respectively, and the target scene images corresponding to the left and right cameras are obtained. Generally speaking, a binocular stereo matching algorithm is used to obtain the depth information of the target scene based on the target scene image, thereby obtaining a depth map corresponding to the target scene. In this embodiment, the target scene image corresponding to the target scene that has been supplemented with light is acquired to obtain an image with richer texture information, thereby improving the accuracy of extracting depth information from the target scene image.

[0043] In one implementation, determining the depth map of the target scene according to the target scene image includes: Step S301, calculating the matching cost between pixels corresponding to the target scene images respectively acquired by the left and right cameras in the binocular camera based on the similarity measurement; Step S302: determining the aggregate matching cost corresponding to each pixel according to the aggregate window and each matching cost; Step S303, performing disparity calculation based on the aggregate matching cost to determine a disparity map; Step S304: Obtain binocular camera parameters, and determine the depth map according to the binocular camera parameters and the disparity map.

[0044] Specifically, before calculating the matching cost between each pixel in the target scene image, the left and right target scene images are preprocessed, such as grayscale, denoising, etc. Considering that different positions on the image have different sets of surrounding pixels and different image region structures, the method for calculating the matching cost between the pixels in the left and right target scene images after preprocessing based on the similarity metric in this embodiment is: taking a fixed-size window (feature window) centered on a pixel, and comparing the image block similarity of the feature windows corresponding to the pixels in the left and right target scene images, thereby obtaining the matching cost of the pixel.

[0045] Since the matching cost takes local information into consideration, the cost value is calculated by calculating the pixel information in a window of a certain size within the pixel area, which is easily affected by image noise. When the image is in a weak texture or repeated texture + area, the matching cost is very likely to fail to accurately reflect the correlation between pixels. This embodiment performs cost aggregation based on the matching cost to obtain the aggregated matching cost corresponding to each pixel. Cost aggregation enables the matching cost to accurately reflect the correlation between pixels by establishing a connection between adjacent pixels. For two pixels in the left and right target scene images, windows of the same size (aggregation windows) are taken with these two pixels as the center, and then the matching costs between the pixels in the same position in the two windows are calculated one by one, and these matching costs are accumulated as the final aggregated matching cost. If the aggregation cost is smaller, the two pixels are more likely to be matched. Cost aggregation methods include: scan line method, dynamic programming method, path aggregation method, etc.

[0046] After the aggregate matching cost is calculated, the optimal disparity value of each pixel can be determined by the aggregate matching cost corresponding to each pixel. In this embodiment, the winner-takes-all algorithm (WTA) is used to calculate the disparity of each pixel, that is, for a pixel in the target scene graph on the left, it has multiple possible matching pixels in the target scene graph on the right, and the aggregate matching cost of these possible matching pixels and the current pixel to be matched are compared one by one, and the pixel with the minimum aggregate matching cost is the matching pixel found.

[0047] After obtaining the disparity map, the disparity map can be converted into a depth map based on the camera parameters, where the conversion formula between disparity and depth is: ; represents the depth map, Represents the normalized focal length, which is the camera internal parameter, is the distance between the optical centers of the two cameras, i.e. the baseline distance, is the parallax value.

[0048] In one implementation, before determining the depth map according to the binocular camera parameters and the disparity map, the method further includes: The disparity map is subjected to disparity post-processing, wherein the disparity post-processing method includes: one or more of left-right consistency detection, disparity filling, and median filtering.

[0049] Specifically, before converting the disparity map into a depth map, the disparity map can be first subjected to disparity post-processing or disparity optimization. Among them, the implementation method of left-right consistency detection is: after the disparity map corresponding to the left target scene image has been obtained, the left and right images are swapped, that is, the left image becomes the right image, and the right image becomes the left image, and then stereo matching is performed again to obtain the disparity map of the new left target scene image (the original right target scene image), and the disparity of the same-name points of the left and right images is compared based on the threshold to see whether the disparity is consistent. If it is consistent, the consistency check is satisfied, otherwise it is not satisfied, and the unsatisfied points are removed. The implementation method of disparity filling is: determine whether the pixel point is an occluded point or a mismatched point. When the pixel point is an occluded point, fill it based on the minimum disparity of the disparity value of the surrounding background pixels; when the pixel point is a mismatched point, fill it based on the median of the surrounding disparity values. The disparity post-processing method adopted in this embodiment includes one or more of left-right consistency detection, disparity filling and median filtering, so as to improve the accuracy of depth map calculation.

[0050] Based on the above embodiments, the present invention also provides a binocular imaging device based on low-light scenes, such as Figure 2 As shown, the device comprises: Parameter determination module 01, used to collect an initial scene image corresponding to an initial scene, and determine illumination parameters according to the initial scene image; The light and dark adjustment module 02 is used to fill the initial scene with light according to the illumination parameters to determine the target scene; The binocular imaging module 03 is used to collect a target scene image corresponding to the target scene based on a binocular camera, and determine a depth map of the target scene according to the target scene image.

[0051] Based on the above embodiment, the present invention further provides a terminal, whose principle block diagram can be shown as follows: Figure 3 As shown. The terminal includes a processor, a memory, a network interface, and a display screen connected through a system bus. Among them, the processor of the terminal is used to provide computing and control capabilities. The memory of the terminal includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the terminal is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a binocular imaging method based on a low-light scene is implemented. The display screen of the terminal can be a liquid crystal display screen or an electronic ink display screen.

[0052] Those skilled in the art will understand that Figure 3The principle block diagram shown in the figure is only a block diagram of a partial structure related to the scheme of the present invention, and does not constitute a limitation on the terminal to which the scheme of the present invention is applied. The specific terminal may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0053] In one implementation, the memory of the terminal stores one or more programs, and is configured to be executed by one or more processors. The one or more programs include instructions for performing a binocular imaging method based on a low-light scene.

[0054] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0055] In summary, the present invention discloses a binocular imaging method, device, terminal and storage medium based on low-light scenes, wherein the method collects an initial scene image corresponding to the initial scene, determines the illumination parameters according to the initial scene image; fills the initial scene with light according to the illumination parameters to determine the target scene; collects the target scene image corresponding to the target scene based on the binocular camera, and determines the depth map of the target scene according to the target scene image. Since the present invention determines the brightness of the initial scene according to the initial scene image, thereby obtaining the illumination parameters, and fills the insufficiently illuminated area in the initial scene according to the illumination parameters, so that the binocular camera can collect high-quality images based on the scene after fill-in, it can effectively solve the problem in the prior art that the surface texture features of the images captured by the left and right cameras are not obvious due to insufficient illumination, thereby reducing the accuracy of the depth calculation.

[0056] It should be understood that the application of the present invention is not limited to the above examples. For ordinary technicians in this field, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A binocular imaging method based on low-light scenes, characterized in that: The method comprises: Collecting an initial scene image corresponding to the initial scene, and determining illumination parameters according to the initial scene image; Filling the initial scene with light according to the illumination parameters to determine a target scene; A target scene image corresponding to the target scene is collected based on a binocular camera, and a depth map of the target scene is determined according to the target scene image.

2. The binocular imaging method based on low-light scenes according to claim 1, characterized in that: The determining of the illumination parameters according to the initial scene image comprises: Determining a brightness distribution matrix according to the initial scene image, wherein the brightness distribution matrix is ​​used to represent the brightness of each position in the initial scene image; Determine the illumination intensity and illumination angle according to the brightness distribution matrix; The illumination parameter is determined according to the illumination intensity and the illumination angle.

3. The binocular imaging method based on low-light scenes according to claim 2, characterized in that: The step of determining a brightness distribution matrix according to the initial scene image comprises: Determine the position information and light intensity information corresponding to each pixel point according to the initial scene image; The brightness distribution matrix is ​​determined according to each of the position information and each of the light intensity information.

4. The binocular imaging method based on low-light scenes according to claim 2, characterized in that: The determining the illumination intensity and illumination angle according to the brightness distribution matrix comprises: Obtaining a preset brightness value, and determining a point to be filled in light corresponding to the initial scene according to the preset brightness value and the brightness distribution matrix; Acquire a preset range, and determine the light intensity according to the preset range and the point to be filled light; The illumination angle is determined according to a center point corresponding to the initial scene image and the light point to be filled.

5. The binocular imaging method based on low-light scenes according to claim 1, characterized in that: Before collecting the target scene image corresponding to the target scene based on the binocular camera, the method further includes: forming a random dot matrix on the target scene through a diffractive optical element.

6. The binocular imaging method based on low-light scenes according to claim 1, characterized in that: Determining the depth map of the target scene according to the target scene image includes: The matching cost between the pixels corresponding to the target scene images respectively acquired by the left and right cameras of the binocular camera is calculated based on the similarity metric; Determine the aggregate matching cost corresponding to each pixel according to the aggregate window and each of the matching costs; Performing disparity calculation based on the aggregate matching cost to determine a disparity map; Obtain binocular camera parameters, and determine the depth map according to the binocular camera parameters and the disparity map.

7. The binocular imaging method based on low-light scenes according to claim 6, characterized in that: Before determining the depth map according to the binocular camera parameters and the disparity map, the method includes: The disparity map is subjected to disparity post-processing, wherein the disparity post-processing method includes: one or more of left-right consistency detection, disparity filling, and median filtering.

8. A binocular imaging device based on low-light scenes, characterized in that: The device comprises: A parameter determination module, used to collect an initial scene image corresponding to the initial scene, and determine the illumination parameters according to the initial scene image; A brightness adjustment module, used to fill in the initial scene with light according to the illumination parameters to determine the target scene; The binocular imaging module is used to collect a target scene image corresponding to the target scene based on a binocular camera, and determine a depth map of the target scene according to the target scene image.

9. A terminal, characterized in that: The terminal includes a memory and one or more processors; the memory stores one or more programs; the program includes instructions for executing the binocular imaging method based on low-light scenes as described in any one of claims 1-7; and the processor is used to execute the program.

10. A computer-readable storage medium having a plurality of instructions stored thereon, characterized in that: The instructions are suitable for being loaded and executed by a processor to implement the steps of the binocular imaging method based on low-light scenes as described in any one of claims 1 to 7 above.

Images