Water depth detection system based on polarized light and electronic equipment

Abstract

The invention discloses a water depth detection system based on polarized light and electronic equipment, and the system comprises a first imaging unit which is used for obtaining a left eye image and a right eye image in a front view range of a vehicle; the parallax calculation unit is used for calculating a first parallax map based on the left eye image and the right eye image; the judgment unit is used for judging whether parallax in a front road area in the first parallax map is abnormal or not, and if yes, a polarization control instruction is generated; the polarization acquisition unit is used for responding to the polarization control instruction and acquiring a left-eye polarized light image and a right-eye polarized light image in a front view range of the vehicle; and the ponding depth determination unit is used for calculating the ponding depth of the ponding area based on the left-eye polarized light image and the right-eye polarized light image. According to the technical scheme, the polarized light is introduced under the condition that the view field and the light intensity are not affected, road waterlogging depth detection is carried out based on the polarized light and an image processing mode, and the accuracy of waterlogging depth detection can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of autonomous driving, and in particular to a water depth detection system based on polarized light and an electronic device. Background Art

[0002] With the continuous development of autonomous driving technology, its applicable scenarios are becoming increasingly diverse. In water-crossing scenarios (such as rainy days and off-road driving), the vehicle's autonomous driving system needs to have a water depth detection function.

[0003] In the existing technology, due to the limitations when processing images containing water surfaces (such as water surface reflections), on the one hand, an additional ultrasonic sensor can be added to the vehicle to calculate the water depth based on the ultrasonic echo; on the other hand, the vehicle can be positioned by a positioning signal (such as a GPS positioning signal, a Beidou positioning signal), and then the water depth can be calculated in combination with the road information. Among them, the first solution undoubtedly increases the cost of the vehicle, and the ultrasonic wave is easily interfered by environmental factors; the second solution requires accurate map information (road height, slope), but for some roads, especially urban and rural roads, there is no accurate map information, and the positioning signal will also have errors, affecting the accuracy of the road height. Therefore, how to detect the depth of water accumulation on roads in different fields is still a problem that needs to be solved.

[0004] Polarized light detection technology can effectively solve the problem of water reflection in images, such as Figure 1 As shown in the figure, applying it in the field of autonomous driving will have the following problems:

[0005] 1) Polarized light cameras at night or in dimly lit environments will weaken the light intensity of light sources such as streetlights and traffic lights, making image recognition more difficult;

[0006] 2) The special structure of the polarized light camera will affect the field of view and create blind spots. Summary of the Invention

[0007] The purpose of this application is to: how to detect the depth of road water accumulation based on polarized light and image processing to improve the accuracy of water accumulation depth detection.

[0008] The technical solution of the first aspect of the present application is: a water depth detection system based on polarized light is provided, the system comprising: a first imaging unit, the first imaging unit is used to obtain a left-eye image and a right-eye image within the field of view in front of a vehicle, wherein the left-eye image and the right-eye image are RGB images; a disparity calculation unit, the disparity calculation unit is used to calculate a first disparity map based on the left-eye image and the right-eye image; a judgment unit, the judgment unit is used to judge whether there is an abnormality in the disparity in the front road area in the first disparity map, and if so, generate a polarization control instruction; a polarization acquisition unit, the polarization acquisition unit is used to respond to the polarization control instruction and obtain a left-eye polarized light image and a right-eye polarized light image within the field of view in front of the vehicle; a water depth determination unit, the water depth determination unit is used to calculate the water depth of the water accumulation area based on the left-eye polarized light image and the right-eye polarized light image.

[0009] In any of the above technical solutions, further, the judgment unit is configured to: when the disparity of the road area ahead in the first disparity map meets a preset judgment condition, determine that the disparity of the road area ahead in the first disparity map is abnormal, wherein the preset judgment condition includes at least one of the following conditions: the first disparity missing ratio in the road area ahead in the first disparity map is greater than or equal to a first preset ratio, the disparity map of the road area ahead in the first disparity map is wavy, and there is at least one target located underground in the first disparity map.

[0010] In any of the above technical solutions, further, the polarization acquisition unit includes: a driving subunit, which is used to respond to polarization control instructions and generate driving instructions; a polarization subunit, which includes at least a fixing device, a polarization plate, a moving device, and a driving device, wherein the polarization plate, the moving device, and the driving device are connected in sequence and installed in the fixing device, and the fixing device is used to install the polarization subunit on the lens of the vehicle-mounted binocular camera, and the driving device is used to respond to the driving instruction and control the moving device to perform reciprocating motion to extend or retract the polarization plate. When the polarization plate is extended, the polarization plate is located in front of the lens.

[0011] In any of the above technical solutions, further, the polarization acquisition unit includes: a polarization calculation subunit, which is used to respond to the polarization control instruction and generate the left eye polarized light image and the right eye polarized light image based on the left eye image, the right eye image and the Stokes parameters.

[0012] In any of the above technical solutions, further, the water depth determination unit is configured to: calculate a second disparity map based on the left-eye polarized light image and the right-eye polarized light image; determine whether there is a water accumulation area in the road area ahead based on the second disparity map, and if so, calculate the fitted road surface plane of the road area ahead based on the second disparity map; calculate the fitted water surface plane of the road area ahead based on the effective disparity in the first disparity map; and calculate the water depth of the water accumulation area based on the fitted water surface plane and the fitted road surface plane.

[0013] In any of the above technical solutions, further, the water depth determination unit is configured to: obtain the measured road surface plane of the road area ahead by querying in the map database based on the positioning signal of the vehicle's current position; correct the fitted road surface plane based on the measured road surface plane, and calculate the water depth of the water accumulation area based on the corrected fitted road surface plane and the fitted water surface plane.

[0014] In any of the above technical solutions, further, based on the second disparity map, it is determined whether there is a water accumulation area in the road area ahead, specifically including: counting the second disparity missing ratio in the road area ahead in the second disparity map; determining whether the second disparity missing ratio is less than a second preset ratio, if so, it is determined that there is a water accumulation area, if not, it is determined that there is no water accumulation area.

[0015] In any of the above technical solutions, further, based on the effective disparity in the first disparity map, the fitted water surface plane of the road area ahead is calculated, specifically including: based on the effective disparity in the first disparity map, obtaining the left boundary and the right boundary of the road surface in the first disparity map; according to preset intervals, obtaining road surface point pairs in the left boundary and the right boundary of the road surface respectively, and generating road surface correction line segments based on the road surface point pairs; based on the road surface correction line segments, calculating the fitted water surface plane of the road area ahead.

[0016] In any of the above technical solutions, further, the water depth determination unit is configured to: obtain ripple information of the water accumulation area based on the left eye image and the right eye image respectively; obtain turbidity information of the water accumulation area based on the left eye polarized light image and the right eye polarized light image respectively; based on the ripple information and the turbidity information, predict the water depth of the water accumulation area according to a preset correspondence relationship, wherein the preset correspondence relationship is determined by the deep learning network through labeled sample images.

[0017] The technical solution of the second aspect of the present application is: to provide an electronic device, on which is provided a water depth detection system based on polarized light as any one of the technical solutions of the first aspect mentioned above.

[0018] The beneficial effects of this application are:

[0019] The technical solution in the present application, when it is determined that there is an abnormality in the parallax in the front road area in the first disparity map, controls the polarization acquisition unit to obtain the corresponding polarized light image, and uses the characteristics of polarized light to predict the water depth in the road water accumulation area. It can not only avoid the problems of "light intensity reduction and blind spots in vision" caused by directly using polarized light, but also creatively combine polarized light and image processing, which helps to improve the accuracy of water depth detection.

[0020] This application also utilizes the property that polarized light can eliminate water surface reflections, and combines it with the principle of parallax calculation. After determining that there is a waterlogged area, the polarized light image is used to calculate the road surface hidden under the water surface, and the road surface height is obtained. The RGB image is then combined to obtain the water surface height, thereby realizing the calculation of the water depth. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The advantages of the above and / or additional aspects of the present application will become apparent and readily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0022] Figure 1 is a schematic diagram of polarized light image comparison in the prior art;

[0023] Figure 2 is another schematic diagram of polarized light image comparison in the prior art;

[0024] Figure 3 is a schematic block diagram of a water depth detection system based on polarized light according to an embodiment of the present application;

[0025] Figure 4 is a schematic diagram of a road surface boundary according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] In order to more clearly understand the above-mentioned objectives, features and advantages of the present application, the present application is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features therein can be combined with each other in the absence of conflict.

[0027] In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the present application may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present application is not limited to the specific embodiments disclosed below.

[0028] Example 1:

[0029] like Figure 3 As shown, this embodiment provides a water depth detection system 100 based on polarized light, which includes:

[0030] The first imaging unit 10 is used to obtain a left-eye image and a right-eye image within the field of view in front of the vehicle, wherein the left-eye image and the right-eye image are RGB images;

[0031] A disparity calculation unit 20 is configured to calculate a first disparity map based on the left-eye image and the right-eye image;

[0032] The judging unit 30 is configured to judge whether there is an abnormality in the parallax of the road ahead in the first disparity map. If so, a polarization control instruction is generated. If not, the first imaging unit 10 and the disparity calculation unit 20 continue to acquire the left-eye image, the right-eye image, and the first disparity map.

[0033] The polarization acquisition unit 40 is used to respond to the polarization control instruction and acquire the left eye polarized light image and the right eye polarized light image within the field of view in front of the vehicle;

[0034] The water depth determination unit 50 is used to calculate the water depth of the water area based on the left polarized light image and the right polarized light image.

[0035] In this embodiment, a binocular camera is installed in front of the autonomous driving vehicle, which includes two left and right cameras installed on a beam. The image data collected by the two cameras is processed (such as distortion correction, cropping, etc.) by the first imaging unit 10 built into the camera, and can output RGB images of a preset resolution, which can be recorded as the left eye image and the right eye image respectively.

[0036] First, a world coordinate system is established with the vertical point of the binocular camera's optical center on the ground as the origin, where the horizontal right direction is the positive x-axis direction, the front of the vehicle is the positive z-axis direction, and the vertical upward direction is the positive y-axis direction. Then, using the disparity calculation principle, the disparity calculation unit 20 calculates the disparity map (first disparity map) corresponding to the left image with the left image as the reference. Then, based on the binocular camera parameters, the position of any target in the left image in the current world coordinate system can be obtained, where the left image can be used for target recognition. Finally, by matching this information one by one, the category of the target in front of the vehicle and the distance between the target and the vehicle can be obtained.

[0037] In this embodiment, the road area ahead of the vehicle during its travel may be considered as a fixed area in the image, such as the middle area in the lower middle portion of the image.

[0038] It should be noted that the disparity calculation process is mainly based on pixel matching of the pixel textures in the left and right images. However, for scenes with reflections from accumulated water, mirror reflections will occur. In addition, the influence of light in different scenes must also be considered. These factors are reflected in the disparity map as missing disparity, false / underground targets, and intermittent ripples. Among them, intermittent ripple-type disparity mainly refers to the disparity caused by fluctuations in the water surface. However, those skilled in the art will understand that images obtained using polarized light can eliminate reflections caused by accumulated water on the road and enhance the image texture.

[0039] Therefore, certain characteristics of road water accumulation in the disparity map can be used as the basis for determining whether road water accumulation has occurred. The judgment unit 30 determines whether the disparity in the forward road area in the first disparity map is abnormal. If these conditions are met, i.e., if the disparity in the forward road area in the first disparity map is determined to be abnormal, water depth detection is performed. The judgment unit 30 generates a polarization control instruction, causing the polarization acquisition unit 40 to obtain left-eye and right-eye polarized light images within the vehicle's forward field of view. Otherwise, conventional disparity calculation and image detection continue. This avoids the aforementioned problems associated with directly using polarized light.

[0040] It should be noted that the required RGB and polarized light images can be obtained by configuring a special image sensor. For example, based on a conventional image sensor, a certain size area (e.g., 2*2) is used as a repeating unit. Within this unit, polarization filters of different angles are set at the corresponding pixels according to a certain pattern. Then, based on the aforementioned polarization control instructions, when a polarized light image is not required, only the data of the pixels not blocked by the polarization filter is obtained to obtain the RGB image; when a polarized light image is required, only the data of the pixels blocked by the polarization filter is obtained to obtain the polarized light image.

[0041] After acquiring the left and right polarized light images, the water depth determination unit 50 calculates disparity to obtain a second disparity map corresponding to the left polarized light image. Based on the road surface fitting method of the disparity map, the road surface plane hidden under the water surface in the current world coordinate system is determined, thereby obtaining the road surface height. This method avoids relying on querying road surface information on a map using the vehicle's positioning signal to obtain "road surface height," helping to solve the problem of "inaccurate positioning signals and missing map information."

[0042] When obtaining the surface of the water, the disparity data of the first disparity map, excluding the water area / disparity abnormal area, can be used to fit the surface of the water in the current world coordinate system, thereby obtaining the water surface height. The difference between the water surface height and the road surface height is the depth of the water.

[0043] It should be noted that since the water surface is a plane, the water surface can be obtained by obtaining its edge in the image and then performing plane fitting to obtain the water surface plane. For the road surface plane, the lowest point of the road surface in the second disparity image can be selected as the correction parameter to determine the potential maximum water depth and avoid danger when the vehicle is wading.

[0044] This embodiment combines polarized light technology with binocular vision technology, and introduces polarized light only when it is determined that there is water accumulation on the road, avoiding the problems of "blind spots and reduced light" caused by using polarized light in normal driving environments.

[0045] In any of the above embodiments, further, in order to avoid frequently starting the polarization acquisition unit 40, the parallax abnormality detection in the front road area in the first parallax map is optimized, and the judgment unit 30 is configured to:

[0046] When the disparity of the road ahead region in the first disparity map satisfies a preset determination condition, it is determined that the disparity of the road ahead region in the first disparity map is abnormal, wherein the preset determination condition includes at least one of the following conditions:

[0047] Considering that when there is glare in the front road area, the pixel texture in the area will be missing in the left and right eye images, so that there is no disparity in this part of the disparity map. Therefore, the number of disparity points with a disparity value of 0 in the front road area in the first disparity map can be counted to obtain a first disparity missing ratio / area, which is used as the above-mentioned preset judgment condition, that is, the first disparity missing ratio in the front road area in the first disparity map is greater than or equal to the first preset ratio.

[0048] Considering that there are multiple vehicles on the road, and the vehicle ahead passing through the flooded area will cause ripples on the water surface, similarly, raindrops falling on the water surface during rain will also cause ripples. Under certain lighting conditions, this part of the "ripples" will serve as image texture and can be reflected in the disparity map. The disparity map corresponding to this situation can also be used as one of the above-mentioned preset judgment conditions, that is, the disparity map of the road ahead in the first disparity map has a ripple / fish-scale distribution.

[0049] Taking into account that under certain lighting conditions, there will be reflections in the waterlogged area, such as mapping the taillights of the vehicle in front, roadside trees, lamp poles, etc., after the corresponding disparity map is projected into a V map, the objects in the V disparity map and the fitted ground can be obtained by clustering and fitting. The "objects" formed by such reflections will be located below the fitted ground and will be judged as false targets. Therefore, such "targets" can also be used as one of the above-mentioned preset judgment conditions, that is, there is at least one target located underground in the first disparity map.

[0050] In any of the above embodiments, further considering the implementation cost and technical difficulty, this embodiment also refers to the shutter structure of a rolling camera and proposes an implementation method of an external polarizing plate / polarizer. By controlling the movement of the polarizing plate, each lens of the binocular camera is blocked to obtain a polarized light image. Therefore, the corresponding polarization acquisition unit 40 includes:

[0051] A driving subunit, the driving subunit is used to respond to the polarization control instruction and generate a driving instruction;

[0052] The polarization subunit includes at least a fixing device, a polarizing plate, a moving device, and a driving device, wherein:

[0053] The polarizing plate, the motion device, and the driving device are connected in sequence and installed in a fixing device. The fixing device is used to install the polarizing subunit on the lens of the vehicle-mounted binocular camera. The fixing device can be set into a cover shape to facilitate installation.

[0054] The driving device is used to respond to the driving instruction and control the motion device to perform a reciprocating motion so as to extend or retract the polarizing plate. When the polarizing plate is extended, the polarizing plate is located in front of the lens.

[0055] Specifically, the driving device can be set to generate a first driving signal after receiving a driving instruction, drive the motion device to move, and translate and extend the polarizing plate so that it blocks the front of the lens. External light passes through the polarizing plate and enters the lens of the vehicle-mounted binocular camera. Then, a conventional image sensor performs photoelectric conversion. After corresponding processing, a polarized light image can be obtained; after that, a delay of a certain time can be set, such as 1 second. After the vehicle-mounted binocular camera has captured several images, the driving device generates a second driving signal to make the motion device move in the opposite direction, retract the polarizing plate, and allow external light to directly enter the lens of the vehicle-mounted binocular camera.

[0056] It should be noted that the transmittance, extinction ratio, acceptance angle and other parameters of the polarizing plate can be set according to actual needs.

[0057] In any of the above embodiments, a deep learning network can be further used to generate a corresponding polarized light image based on an optical algorithm using RGB images and Stokes parameters. Therefore, the polarization acquisition unit 40 further includes:

[0058] The polarization calculation subunit is used to respond to the polarization control instruction and generate the left eye polarized light image and the right eye polarized light image based on the left eye image, the right eye image and the Stokes parameters.

[0059] In any of the above embodiments, further, considering that the factor causing the first disparity map to meet the preset judgment condition is not necessarily waterlogging, but may also be other situations, such as: a "dead end" on a rural road causing a large area of ​​road disparity loss; the influence of strong lighting on an urban road; or the projection lights of a large billboard at night. Therefore, it is necessary to perform a secondary judgment on whether there is waterlogging in the road ahead, and the water depth determination unit 50 is configured to:

[0060] Calculating a second disparity map based on the left-eye polarized light image and the right-eye polarized light image;

[0061] Based on the second disparity map, determine whether there is any water accumulation in the road ahead. That is, whether the disparity in the area of ​​the second disparity map with the disparity deviation from the first disparity map has fully or partially returned to normal (the disparity value is not 0). If so, calculate the fitted road surface plane of the road ahead using a plane fitting method based on the second disparity map. If not, it indicates that the anomaly in the first disparity map is not caused by water accumulation, and in this case, no subsequent water depth detection is required.

[0062] Based on the effective disparity in the first disparity map, a fitted water surface plane of the front road area is calculated; wherein the effective disparity is the disparity in the first disparity map excluding the abnormal disparity area in the front road area, and can be located within the front road area or outside the front road area.

[0063] Those skilled in the art will appreciate that image recognition can be used to identify objects on both sides of the road in the left-view image, such as curbs / shoulders, railings, plants, etc. These objects will correspond to a target frame in the left-view image, which will contain location information in the image (pixel coordinates of the target frame). Since the left-view image corresponds to the disparity map, the image position of such objects in the first disparity map can be obtained, and thus their corresponding disparity values.

[0064] For example, when fitting a water surface, the disparity values ​​of the bottom of the curbs and railings on both sides of the road can be selected as the data for fitting the water surface. During the fitting process, the disparity values ​​of these targets can be converted into 3D point cloud data in the current world coordinate system based on the parameters (intrinsic and extrinsic) of the onboard binocular camera. Combined with the plane fitting method, the fitted water surface is obtained.

[0065] Based on the fitted water surface plane and the fitted road surface plane, the water depth of the flooded area can be calculated, that is, the difference in the y-axis heights of the two planes in the current coordinate system.

[0066] In any of the above embodiments, further, based on the second disparity map, determining whether there is a waterlogged area in the road area ahead specifically includes:

[0067] Considering the complex road conditions and the presence of disparity holes, especially in non-dense disparity maps, after obtaining the second disparity map calculated based on the polarized light image, the second disparity missing ratio in the forward road area in the second disparity map is calculated, that is, the proportion / ratio of points with a disparity value of 0.

[0068] It is determined whether the second disparity missing ratio in the front road area in the second disparity map is less than a second preset ratio. If so, it is determined that a water accumulation area exists; if not, it is determined that no water accumulation area exists.

[0069] In any of the above embodiments, further, in order to improve the accuracy of the fitted road surface plane, for vehicles traveling in cities, the fitted road surface plane can be corrected using road surface information in a map obtained through positioning signals. The water depth determination unit 50 is configured as follows:

[0070] Based on the positioning signal of the vehicle's current position, the measured road surface plane of the road area ahead is obtained by querying the map database; if no positioning signal is obtained or the measured road surface is not obtained, the road surface correction process is skipped.

[0071] Based on the measured road surface, the fitted road surface is corrected, and the water depth of the flooded area is calculated based on the corrected fitted road surface and the fitted water surface. When correcting the fitted road surface, the difference between the two can be calculated to avoid problems caused by inaccurate positioning signals or map data. When the difference is less than a preset threshold, the fitted road surface can be corrected according to a set weighting factor, such as setting the weight of the measured road surface to 20% and the weight of the original fitted road surface to 80%.

[0072] In any of the above embodiments, further, Figure 4 As shown, considering that the water surface plane is not affected by the terrain, the fitted water surface plane can be corrected based on the boundary of the water surface, and the line between two points on the boundary should be located on the plane where the water surface is located. Therefore, the effective disparity is set to be the disparity in the first disparity map excluding the disparity abnormal area in the front road area. Based on the effective disparity in the first disparity map, the fitted water surface plane of the front road area is calculated, which specifically includes:

[0073] Based on the effective disparity in the first disparity map, the left boundary and the right boundary of the road surface in the first disparity map are obtained; that is, Figure 4 Regarding the middle curve part, it should be noted that the boundary can also be set to a portion of the curve within a certain range, such as taking 20% ​​of the entire width from the outermost side as the obtained road surface boundary;

[0074] At preset intervals, for example, with a maximum step size of 5 pixels / parallax point, randomly obtain road point pairs (A, B) at the left and right edges of the road surface, and generate road correction line segments based on these road point pairs. The road point pairs (A, B) can be connected to form a straight line, which should lie on the fitted water surface plane.

[0075] Specifically, you can randomly select point A on the left boundary, and then randomly select a step size within the preset interval, such as 3, and then move the row coordinate of point A up / down by 3 pixels. The corresponding point B on the right boundary is the selected point, and forms a road point pair (A, B) with point A.

[0076] Based on the road surface correction line segment, the fitted water surface plane of the road area ahead is calculated.

[0077] Example 2:

[0078] Considering that polarized light technology can not only eliminate reflections on the water surface, but also detect the concentration of liquid / gas through polarized light rotation and polarized light imaging methods, the concentration / turbidity of waterlogged areas on the road can be predicted based on polarized light images combined with deep learning networks.

[0079] Based on the above embodiment, this embodiment introduces a deep learning network to perform network training to obtain the corresponding relationship between water depth, water turbidity, and water surface ripples, thereby predicting the road water depth based on the acquired RGB image and polarized light image. Therefore, the water depth determination unit 50 is configured as follows:

[0080] Based on the left image and the right image respectively, the ripple information of the water accumulation area is obtained;

[0081] Obtain turbidity information of the waterlogged area based on the left-eye polarized light image and the right-eye polarized light image respectively;

[0082] Based on the ripple information and turbidity information, the water depth in the waterlogged area is predicted according to the preset correspondence relationship, wherein the preset correspondence relationship is determined by the deep learning network through labeled sample images.

[0083] Specifically, for waterlogging on urban roads, the following conditions can be set:

[0084] 1. Under the same rainfall conditions, if the water is deeper, the ripples produced by raindrops falling on the water surface are larger and the splashes are smaller; conversely, the ripples are smaller and the splashes are larger;

[0085] 2. When considering the flow of water on the road, if the water is deeper, the sand content in the water is higher and the water is more turbid; conversely, the sand content is lower and the water is clearer.

[0086] Based on the above-mentioned conditions, the collected data is annotated, and the water depth corresponding to the data collection is annotated as a known data, and a database of the deep learning network is constructed for training. It enables it to determine the (preset) correspondence between the size of the water surface ripples and the turbidity of the accumulated water in the image and the depth of the accumulated water, so as to realize the prediction of the accumulated water depth based on the collected image.

[0087] Considering that the onboard binocular camera can capture left and right eye images (RGB) and left and right eye polarized images, these images are fed into the deep learning network to improve its accuracy in predicting water depth. The left and right eye images are used to determine the size of water surface ripples, which serves as ripple information; the left and right polarized images are used to determine the turbidity of the water, which serves as turbidity information. Using these ripple and turbidity information, the deep learning network can then predict the water depth of the flooded area according to the pre-set correspondence relationship established after training.

[0088] Example 3:

[0089] This embodiment further provides an electronic device, which is provided with any of the polarized light-based water depth detection systems described in the above embodiments. The electronic device may include a processor, a memory, an input / output interface, and a communication interface. The processor, the memory, and the input / output interface are connected via a system bus, and the communication interface, the data acquisition unit, and the input device are connected to the system bus via the input / output interface.

[0090] It should be noted that the composition structure of the above electronic equipment can be adjusted according to actual needs.

[0091] So far, the various embodiments of the present application have been described in detail. To avoid obscuring the concept of the present application, some details well known in the art have not been described. Based on the above description, those skilled in the art can fully understand how to implement the technical solutions disclosed herein.

[0092] Although some specific embodiments of the present application have been described in detail through examples, those skilled in the art should understand that the above examples are only for illustration and are not intended to limit the scope of the present application.

[0093] The steps in this application can be adjusted in order, combined, and deleted according to actual needs.

[0094] Although the present application is disclosed in detail with reference to the accompanying drawings, it should be understood that these descriptions are merely exemplary and are not intended to limit the application of the present application. The scope of protection of the present application is defined by the appended claims and may include various modifications, alterations and equivalents made to the invention without departing from the scope and spirit of the present application.

Claims

1. A water depth detection system based on polarized light, characterized in that: The system comprises: A first imaging unit, the first imaging unit is used to obtain a left-eye image and a right-eye image within a field of view in front of the vehicle, wherein the left-eye image and the right-eye image are RGB images; A disparity calculation unit, configured to calculate a first disparity map based on the left-eye image and the right-eye image; A judging unit, the judging unit being used to judge whether there is an abnormality in the parallax in the front road area in the first parallax map, and if so, generate a polarization control instruction; A polarization acquisition unit, the polarization acquisition unit is used to respond to the polarization control instruction and acquire a left-eye polarized light image and a right-eye polarized light image within the field of view in front of the vehicle; A water accumulation depth determination unit is used to calculate the water accumulation depth of the water accumulation area based on the left-eye polarized light image and the right-eye polarized light image.

2. The water depth detection system based on polarized light according to claim 1, characterized in that: The judging unit is configured to: When the disparity of the road area ahead in the first disparity map satisfies a preset determination condition, it is determined that the disparity of the road area ahead in the first disparity map is abnormal, wherein the preset determination condition includes at least one of the following conditions: The first disparity missing ratio in the front road area in the first disparity map is greater than or equal to a first preset ratio, The disparity map of the road area ahead in the first disparity map is wavy. There is at least one target located underground in the first disparity map.

3. The water depth detection system based on polarized light according to claim 1, characterized in that: The polarization acquisition unit comprises: A driving subunit, the driving subunit being used to generate a driving instruction in response to the polarization control instruction; A polarization subunit, the polarization subunit at least comprises a fixing device, a polarization plate, a moving device, and a driving device, wherein: The polarizing plate, the moving device, and the driving device are connected in sequence and installed in the fixing device, and the fixing device is used to install the polarizing subunit on the lens of the vehicle-mounted binocular camera. The driving device is used to respond to the driving instruction and control the moving device to perform a reciprocating motion so as to extend or retract the polarizing plate. When the polarizing plate is extended, the polarizing plate is located in front of the lens.

4. The water depth detection system based on polarized light according to claim 1, characterized in that: The polarization acquisition unit comprises: A polarization calculation subunit, the polarization calculation subunit is used to respond to the polarization control instruction and generate the left-eye polarized light image and the right-eye polarized light image based on the left-eye image, the right-eye image and Stokes parameters.

5. The water depth detection system based on polarized light according to claim 1, characterized in that: The water depth determination unit is configured to: Calculating a second disparity map based on the left-eye polarized light image and the right-eye polarized light image; Based on the second disparity map, determining whether there is a water accumulation area in the road area ahead, and if so, calculating a fitted road surface plane of the road area ahead based on the second disparity map; Calculating a fitted water surface plane of the road area ahead based on the effective disparity in the first disparity map; The water depth of the water accumulation area is calculated based on the fitted water surface plane and the fitted road surface plane.

6. The water depth detection system based on polarized light according to claim 5, characterized in that: The water depth determination unit is configured to: Based on the positioning signal of the current position of the vehicle, obtaining the measured road surface plane of the road area ahead by querying in a map database; Based on the measured road surface plane, the fitted road surface plane is corrected, and based on the corrected fitted road surface plane and the fitted water surface plane, the water depth of the waterlogged area is calculated.

7. The water depth detection system based on polarized light according to claim 5 or 6, characterized in that: The determining, based on the second disparity map, whether there is a waterlogged area in the road area ahead specifically includes: Counting the missing ratio of the second disparity in the front road area in the second disparity map; It is determined whether the second parallax missing ratio is less than a second preset ratio. If so, it is determined that the water accumulation area exists; if not, it is determined that the water accumulation area does not exist.

8. The water depth detection system based on polarized light according to claim 5, characterized in that: The calculating the fitted water surface plane of the front road area based on the effective disparity in the first disparity map specifically includes: Based on the effective disparity in the first disparity map, obtaining a left boundary and a right boundary of a road surface in the first disparity map; Acquire road surface point pairs at the left boundary of the road surface and the right boundary of the road surface respectively at preset intervals, and generate road surface correction line segments based on the road surface point pairs; Based on the road surface correction line segment, a fitting water surface plane of the front road area is calculated.

9. The water depth detection system based on polarized light according to any one of claims 1 to 4, characterized in that: The water depth determination unit is configured to: Based on the left-eye image and the right-eye image respectively, acquiring ripple information of the water accumulation area; Based on the left-eye polarized light image and the right-eye polarized light image, respectively, obtaining turbidity information of the waterlogged area; Based on the ripple information and the turbidity information, the water depth in the waterlogged area is predicted according to a preset corresponding relationship, wherein the preset corresponding relationship is determined by a deep learning network through labeled sample images.

10. An electronic device, characterized in that: The electronic device is provided with a water depth detection system based on polarized light as claimed in any one of claims 1 to 9.

Images