Object detection method and device

Abstract

The present application provides an object detection method and device, wherein the object detection method includes: obtaining a to-be-processed disparity image and a regional feature disparity image for a to-be-recognized area, wherein the to-be-recognized area includes at least two detection sub-areas; comparing the to-be-processed disparity image and the regional feature disparity image based on the disparity value feature information of the pixel points in each detection sub-area, and determining the abnormal pixel points in each detection sub-area; determining the comparison value corresponding to each detection sub-area according to the abnormal pixel points in each detection sub-area, and generating the comparison result of the to-be-processed disparity image according to the comparison value corresponding to each detection sub-area; and determining whether there is a target object in the to-be-recognized area according to the comparison result. Through this method, a disparity image is used to detect whether there is a target object in the to-be-recognized area, and the processing of the disparity value in the disparity image occupies less data resources, and the data processing speed is fast and efficient, which can effectively improve the efficiency of the object detection method.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an object detection method, an object detection device, a computing device, a computer-readable storage medium, and a computer program product. Background Art

[0002] Currently, using machine vision for scene recognition in public places has become a hot research field. More and more researchers are trying to apply deep learning, computer vision and other technologies to intelligent transportation, security monitoring and other fields. In urban rail transit, gate passage judgment is an important application. For example, in a gate passage scenario, a passenger needs to swipe a card to authorize before entering the station. After the authorization is passed, the gate door opens and closes automatically after the passenger passes.

[0003] Traditionally, monitoring of people passing through the gate is achieved by a series of infrared beam sensors installed on the inside of the gate, assisted by diffuse reflection sensors, which can sense the entry and exit status of people and control the gate's opening and closing actions at the appropriate time. Some new smart gates now use machine vision technology to replace traditional infrared sensor solutions, because the camera has a larger field of view and richer scene information, which can assist the algorithm to achieve more content recognition. However, the passenger recognition frame obtained by the RGB image of the top camera alone in conjunction with the image detection and tracking algorithm may be lost, which to a certain extent will interfere with the gate's passage software's judgment of the passenger's passage status, resulting in the inability to recognize that the passenger has completed the passage, and the gate door cannot be closed in time or a scene where the gate door shoots people. Summary of the invention

[0004] In view of this, the embodiment of the present application provides an object detection method. The present application also relates to an object detection device, a computing device, a computer-readable storage medium and a computer program product to solve the above problems existing in the prior art.

[0005] According to a first aspect of an embodiment of the present application, there is provided an object detection method, comprising:

[0006] Acquire a to-be-processed disparity image and a regional feature disparity image for a region to be identified, wherein the region to be identified includes at least two detection sub-regions;

[0007] Based on the disparity value feature information of the pixel points in each detection sub-region, the disparity image to be processed is compared with the regional feature disparity image to determine the abnormal pixel points in each detection sub-region;

[0008] Determining a comparison value corresponding to each detection sub-region according to abnormal pixel points in each detection sub-region, and generating a comparison result of the parallax image to be processed according to the comparison value corresponding to each detection sub-region;

[0009] Determine whether there is a target object in the area to be identified according to the comparison result.

[0010] According to a second aspect of an embodiment of the present application, there is provided an object detection device, including:

[0011] An acquisition module is configured to acquire a to-be-processed disparity image and a regional feature disparity image for a region to be identified, wherein the region to be identified includes at least two detection sub-regions;

[0012] A determination module is configured to compare the to-be-processed disparity image with the regional characteristic disparity image based on the disparity value characteristic information of the pixel points in each detection sub-region, and determine the abnormal pixel points in each detection sub-region;

[0013] A generating module, configured to determine a comparison value corresponding to each detection sub-region according to abnormal pixel points in each detection sub-region, and generate a comparison result of the parallax image to be processed according to the comparison value corresponding to each detection sub-region;

[0014] The detection module is configured to determine whether there is a target object in the area to be identified according to the comparison result.

[0015] According to a third aspect of an embodiment of the present application, a computing device is provided, including:

[0016] Memory and processor;

[0017] The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions. When the computer programs / instructions are executed by the processor, the steps of the above method are implemented.

[0018] According to a fourth aspect of an embodiment of the present application, a computer-readable storage medium is provided, which stores a computer program / instruction, and the steps of the above method are implemented when the computer program / instruction is executed by a processor.

[0019] According to a fifth aspect of an embodiment of the present application, a computer program product is provided, comprising a computer program / instruction, which implements the steps of the above method when executed by a processor.

[0020] The object detection method provided by the present application obtains a disparity image to be processed and a regional feature disparity image for a region to be identified, wherein the region to be identified includes at least two detection sub-regions; compares the disparity image to be processed and the regional feature disparity image based on the disparity value feature information of the pixel points in each detection sub-region, and determines the abnormal pixel points in each detection sub-region; determines the comparison value corresponding to each detection sub-region according to the abnormal pixel points in each detection sub-region, and generates the comparison result of the disparity image to be processed according to the comparison value corresponding to each detection sub-region; and determines whether there is a target object in the region to be identified according to the comparison result.

[0021] An object detection method provided in an embodiment of the present application adopts a disparity image method to detect whether there is a target object in the identification area. The processing of the disparity value in the disparity image occupies less data resources, has a fast data processing speed and high efficiency, and can effectively improve the efficiency of the object detection method. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 is a flow chart of an object detection method provided by an embodiment of the present application;

[0023] Figure 2 is a schematic diagram of an area to be identified and a detection sub-area provided in an embodiment of the present application;

[0024] Figure 3 It is a processing flow chart of an object detection method applied to a gate switch control scene provided by an embodiment of the present application;

[0025] Figure 4 It is a schematic diagram of a passenger passing through a gate outside a safety zone provided by an embodiment of the present application;

[0026] Figure 5 is a structural schematic diagram of an object detection device provided by an embodiment of the present application;

[0027] Figure 6 It is a structural block diagram of a computing device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0028] Many specific details are described in the following description to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present application, so the present application is not limited by the specific implementation disclosed below.

[0029] The terms used in one or more embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit one or more embodiments of the present application. The singular forms of "a", "said" and "the" used in one or more embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in one or more embodiments of the present application refers to and includes any or all possible combinations of one or more associated listed items.

[0030] It should be understood that, although the terms first, second, etc. may be used to describe various information in one or more embodiments of the present application, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of the present application, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0031] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards in the relevant regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0032] First, the terms involved in one or more embodiments of the present application are explained.

[0033] Disparity image: A disparity image refers to a two-dimensional image that stores the disparity values ​​of all pixels in the image after stereo correction. A disparity image is a two-dimensional image whose size is the same as the original image. Each position in the disparity image stores the disparity characteristics of the pixels at that position in pixels.

[0034] Structured light camera: a 3D imaging device based on the principle of active stereo vision, which uses structured light technology to obtain three-dimensional geometric information of objects.

[0035] RGB image: It is a color image representation method based on the combination of three color channels: red, green, and blue. In an RGB image, the color of each pixel is determined by the intensity values ​​of these three channels, and the intensity range of each channel is usually 0 to 255. In digital image processing and computer graphics, the data of RGB images is usually organized into a three-dimensional array, where two dimensions represent the width and height of the image, and the third dimension represents the three color channels of red, green, and blue.

[0036] Currently, using machine vision for scene recognition in public places has become a hot research field. More and more researchers are trying to apply machine learning, computer vision and other technologies to intelligent transportation, security monitoring and other fields. In urban rail transit, the implementation of gate passage logic is an important application. A simple scenario is that passengers need to swipe their cards for authorization before entering the station. After the authorization is passed, the gate door opens and closes automatically after the passengers pass through.

[0037] The personnel judgment of traditional gates is realized by a series of infrared counter-radiation sensors installed on the inside of the gates, assisted by diffuse reflection sensors, which can sense the entry and exit status of personnel and control the opening and closing of the gates at the appropriate time. Some of the existing new smart gates use machine vision technology to replace the traditional infrared sensor solution. Because the camera has a larger field of view and richer scene information, it can assist the algorithm to realize more content recognition. However, the passenger recognition frame obtained by the RGB image of the top camera alone with the image detection and tracking algorithm may be lost, which will interfere with the judgment of the passenger communication status of the gate pass software to a certain extent, resulting in the inability to recognize that the passenger has passed and the gate door cannot be closed in time or the gate door shoots people.

[0038] Based on this, in the present application, an object detection method is provided. The present application also relates to an object detection device, a computing device, a computer-readable storage medium and a computer program product, which are described in detail one by one in the following embodiments.

[0039] Figure 1 A flow chart of an object detection method provided according to an embodiment of the present application is shown, which specifically includes the following steps:

[0040] Step 102: Obtain a to-be-processed disparity image and a regional feature disparity image for a region to be identified, wherein the region to be identified includes at least two detection sub-regions.

[0041] The object can be understood as an object that exists in the area to be identified. In practical applications, the object can be a biological organism such as a person, a dog, or a cat, or an object such as a box or a package. In the method provided in this application, the specific form of the object is not limited, and the actual application shall prevail.

[0042] The area to be identified can be understood as the area where object detection is required in the embodiment of the present application, that is, the method provided by the present application needs to detect whether there is an object in the area to be identified. In actual applications, the area to be identified can be a gate for people to pass through, a security gate at the entrance and exit of a station, or an X-ray security inspection machine for luggage security inspection, etc. In the present application, the specific form of the area to be identified is not limited, and the actual application shall prevail.

[0043] In the method provided in the present application, the area to be identified includes at least two detection sub-areas, and the detection sub-areas can be understood as areas that need to be focused on to detect whether an object exists during the object detection process. For example, when the area to be identified is a gate, the detection sub-area can be understood as the area distributed before and after the gate; when the area to be identified is an X-ray security inspection machine, the inspection sub-area can be the area of ​​the entrance and exit of the X-ray security inspection machine. In practical applications, in the same area to be identified, the more detection sub-areas there are, the higher the detection accuracy, and the detection sub-areas will not overlap.

[0044] like Figure 2 As shown, Figure 2 A schematic diagram of a region to be identified and a detection sub-region provided in an embodiment of the present application is shown. Figure 2 As shown, the area where pedestrians walk is the area to be identified, and two detection sub-areas are set in the area to be identified. The object detection method provided in the present application is specifically a method for detecting whether there is an object in the detection sub-area.

[0045] The parallax image to be processed can be understood as a parallax image acquired by an image acquisition device for the area to be identified. In the method provided in the present application, the image acquisition device can be a structured light camera, the image acquisition device is arranged above the area to be identified, the lens of the image acquisition device is aimed at the area to be identified, and the area to be identified can be photographed from a bird's-eye view.

[0046] A structured light camera is a camera that uses structured light for three-dimensional imaging. It mainly projects structured light of a known pattern onto the surface of the object to be measured through a projector, and then the structured light camera captures the structured light image of the object surface from multiple angles (usually two angles, forming a binocular system). The captured structured light image is preprocessed, including image denoising, enhancement, and distortion correction, to improve image quality and reduce subsequent calculation errors. It is also necessary to perform epipolar correction on the images at two angles of the binocular system so that the corresponding pixels are on the same horizontal line. Based on the principle of binocular stereo vision, the disparity value is calculated by matching the positions of the corresponding pixels in the left and right images, and finally the corresponding disparity image is generated.

[0047] The regional characteristic disparity image can be understood as a disparity image of the area to be identified when there is no target object. For example, when the area to be identified is a gate, the regional characteristic disparity image can be understood as a disparity characteristic image in a static state acquired by an image acquisition device when no one passes through the gate.

[0048] In the method provided in the present application, a disparity image comparison method is used to detect whether there is an object in each detection sub-area, and then further determine whether there is an object in the disparity image to be processed. Therefore, in addition to obtaining the disparity image to be processed, it is also necessary to obtain the regional feature disparity image of the area to be identified. It should be noted that the regional feature disparity image can be pre-generated and directly obtained during the disparity image comparison process, thereby saving computing and processing resources and improving data processing efficiency.

[0049] In practical applications, the regional feature disparity image can be a disparity image that confirms that there is no object, and the disparity image is used as the regional feature disparity image. However, due to various reasons such as weather and lighting, only one disparity image may cause errors in the final detection result. Therefore, preferably, in a specific embodiment provided in the present application, regional features are fused based on multiple static regional disparity images in the area to be identified, thereby generating a regional feature disparity image of the area to be identified.

[0050] Based on this, obtaining a regional feature disparity image for the area to be identified includes:

[0051] Acquire a plurality of static area disparity images for the area to be identified, wherein the static area disparity images are images in which no target object exists in the area to be identified;

[0052] The pixel feature information of two detection sub-regions in each of the multiple static regional disparity images is fused to generate a regional feature disparity image.

[0053] Among them, the static area disparity image can be understood as a disparity image captured by an image acquisition device when there is no object in the area to be identified. In this embodiment, multiple static area disparity images may include disparity images captured by an image acquisition device at different time points and under different lighting conditions; multiple static area disparity images may also be multiple continuous disparity images within a certain time period. In the method provided in the embodiment of the present application, there is no limitation on the number and acquisition time of multiple static area disparity images, which shall be subject to actual application.

[0054] In practical applications, obtaining a static area disparity image may also include obtaining an initial static area disparity image, and preprocessing the initial static area disparity image to obtain the static area disparity image. The initial static area disparity image may be understood as an original disparity image acquired by an image acquisition device. Preprocessing the initial static area disparity image may include using a Gaussian filter to remove noise from the initial static area disparity image, and the size of the Gaussian filter kernel may be selected according to the size of the initial static area disparity image and the size of the acquisition noise.

[0055] After acquiring multiple static area disparity images, the multiple static area disparity images can be fused. Specifically, the pixel feature information (disparity value feature information) of the pixel points corresponding to each detection sub-area in the multiple static area disparity images is fused to generate a regional feature disparity image.

[0056] Specifically, the pixel feature information of two detection sub-regions in each of the multiple static regional disparity images is fused to generate a regional feature disparity image, including:

[0057] Determine a target pixel point in the target detection sub-region and a reference pixel point corresponding to the target pixel point;

[0058] The target disparity value of the target pixel point and the reference disparity value of each reference pixel point in each static area disparity image;

[0059] Determine pixel feature information of the target pixel point according to each target disparity value and each reference disparity value;

[0060] A regional feature disparity image is generated according to the pixel feature information of each target pixel in each detection sub-region.

[0061] In this embodiment, it is necessary to fuse each pixel in the static area parallax image. Furthermore, in order to improve data processing efficiency and reduce consumption of computing resources, only the pixel in the detection sub-area can be fused. Specifically, in this embodiment, a pixel in the target detection sub-area is used as an example for explanation. The target detection sub-area can be any one of at least two detection sub-areas, and the target pixel can be any pixel in the target detection sub-area.

[0062] First, determine the target pixel point in the target detection area and the reference pixel point corresponding to the target pixel point. In practical applications, the number of reference pixel points is multiple. Furthermore, the target pixel point can be taken as the center, and the pixels within a preset range can be used as the reference pixel points of the target pixel point. For example, the preset range is a 5*5 pixel range. After determining the target pixel point, the reference pixel point corresponding to the target pixel point is determined with the target pixel point as the center and the 5*5 pixel range.

[0063] Since the size and position of each static area disparity image are the same, after determining the target pixel and the reference pixel, the target disparity value corresponding to the target pixel and the reference disparity value corresponding to each reference pixel can be obtained in each static area disparity image. Among them, the target disparity value refers to the disparity value of the target pixel in the static area disparity image, and the reference disparity value is the disparity value of the reference pixel in the static area disparity image. Since there are multiple static area disparity images, there are multiple target disparity values ​​and multiple reference disparity values.

[0064] After determining each target disparity value and each reference disparity value, each target disparity value and each reference disparity value can be fused to determine the pixel feature information of the target pixel. The pixel feature information can be specifically understood as the disparity value feature information of the target pixel after fusing multiple target disparity values ​​and multiple reference disparity values.

[0065] In a specific implementation provided in the present application, determining the pixel feature information of the target pixel point according to each target disparity value and each reference disparity value includes:

[0066] Determining a target disparity mean and a target disparity variance corresponding to the target pixel point according to the target disparity value of the target pixel point in each static area disparity image;

[0067] According to the disparity values ​​of the reference pixels in the disparity images of the static regions, the initial reference disparity mean values ​​corresponding to the reference pixels are calculated, and according to the initial reference disparity mean values, the reference disparity mean values ​​corresponding to the target pixels are calculated;

[0068] The pixel feature information of the target pixel point is determined according to the target disparity mean, the target disparity variance and the reference disparity mean.

[0069] In this embodiment, the calculation method of the pixel feature information of the target pixel point is taken as an example to further explain. After the target disparity value of the target pixel point in each static area disparity image is obtained, the target disparity mean and target disparity variance corresponding to the target pixel point can be calculated. Specifically, refer to the following formula 1 and formula 2.

[0070] Formula 1

[0071] Formula 2

[0072] in, represents the mean target disparity, represents the target disparity value of the target pixel in the i-th static area disparity image, n represents the number of static area disparity images, represents the target disparity variance.

[0073] The target disparity mean and target disparity variance corresponding to each target pixel point can be calculated by the above formula 1 and formula 2.

[0074] The pixel feature information corresponding to the target pixel point includes the reference disparity mean of the reference pixel point corresponding to the target pixel point in addition to the disparity variance of the target disparity mean table. Specifically, for each reference pixel point, according to the calculation method of formula 1, the initial reference disparity mean corresponding to each reference pixel point is first calculated, and then the initial reference disparity mean corresponding to each reference pixel point is added and averaged to calculate the reference disparity mean of the reference pixel point corresponding to the target pixel point.

[0075] At this point, the target disparity mean, target disparity variance and reference disparity mean corresponding to the target pixel are taken as pixel feature information of the target pixel and saved in the regional feature disparity image.

[0076] Step 104: comparing the to-be-processed disparity image with the regional characteristic disparity image based on the disparity value characteristic information of the pixel points in each detection sub-region, and determining the abnormal pixel points in each detection sub-region.

[0077] After obtaining the disparity image to be processed and the regional feature disparity image, it is necessary to compare the detection sub-regions in the image. For example, taking the detection sub-region 1 and the detection sub-region 2 in the region to be identified as an example for explanation, it is necessary to compare the features of the detection sub-region 1 in the disparity image to be processed and the detection sub-region 1 in the regional feature disparity image, and to compare the features of the detection sub-region 2 in the disparity image to be processed and the detection sub-region 2 in the regional feature disparity image. By comparing the disparity value feature information of the pixel points in each detection sub-region, the abnormal pixel points in each detection sub-region are determined.

[0078] An abnormal pixel point can be understood as a pixel point whose disparity value feature information in the to-be-processed disparity image and the regional characteristic disparity image is different. If the disparity value feature information of a certain pixel point in the two regions is inconsistent, it means that the to-be-processed disparity image and the regional characteristic disparity image are different at the pixel point, and the pixel point can be determined to be an abnormal pixel point.

[0079] In the comparison process, the disparity feature information of the pixel points in each detection sub-region is used, and the abnormal pixel points are determined by comparing the disparity feature information of each pixel point in the same detection sub-region in the disparity image to be processed and the regional feature disparity image. In a specific embodiment provided in the present application, the disparity image to be processed and the regional feature disparity image are compared based on the disparity value feature information of the pixel points in each detection sub-region to determine the abnormal pixel points in each detection sub-region, including:

[0080] Determine, according to the target detection sub-region, a region to be compared in the to-be-processed disparity image and a reference comparison region in the regional feature disparity image;

[0081] Abnormal pixels are determined according to the disparity value feature information of each pixel in the area to be compared and the reference comparison disparity feature of each pixel in the reference comparison area.

[0082] In this embodiment, the comparison method of one detection sub-region is taken as an example for explanation, that is, a target detection sub-region is selected from each detection sub-region, and the target detection sub-region can be any one of the detection sub-regions.

[0083] After the target detection sub-region is determined, the area to be compared can be determined in the disparity image to be processed based on the target detection sub-region, and the reference comparison area can be determined in the regional feature disparity image. In other words, the same detection sub-region should be selected in the disparity image to be processed and the regional feature disparity image for subsequent comparison. Comparing only the feature information in the target detection sub-region can reduce the amount of data calculation and improve data processing efficiency.

[0084] After determining the area to be compared and the reference comparison area, the disparity value feature information of the pixels in the two areas shall be compared in turn. If the disparity value feature information of the same pixel in the two areas is inconsistent, it means that the disparity image to be processed and the regional feature disparity image at the pixel are different, and the pixel can be determined to be an abnormal pixel.

[0085] In a specific implementation manner provided in the present application, a further explanation is given on how to determine abnormal pixels. According to the disparity value feature information of each pixel in the to-be-compared area and the reference comparison disparity feature of each pixel in the reference comparison area, the abnormal pixels are determined, including:

[0086] Determine a target pixel and at least one reference pixel corresponding to the target pixel, wherein the target pixel is any pixel in the area to be compared, and each reference pixel is a pixel in the reference area corresponding to the target pixel;

[0087] Obtaining a disparity value to be compared of the target pixel in the area to be compared and an average disparity value to be compared between each reference pixel;

[0088] Obtaining a target disparity mean, a target disparity variance, and a reference disparity mean of a target pixel point in the reference comparison area;

[0089] Determine the disparity feature value interval of the target pixel point according to the target disparity mean and the target disparity variance, and determine the reference comparison disparity value interval corresponding to the target pixel point according to the reference disparity mean and a preset error threshold;

[0090] When the to-be-compared disparity value is not within the disparity characteristic value interval, or the to-be-compared disparity mean value is not within the reference comparison disparity value interval, it is determined that the target pixel is an abnormal pixel.

[0091] In the method provided in the embodiment of the present application, a target pixel point is determined in the area to be compared, and the target pixel point is used as an example to explain whether the target pixel point is an abnormal pixel point. The target pixel point is any pixel point in the area to be compared.

[0092] After determining the target pixel, at least one reference pixel corresponding to the target pixel can be further determined. The reference pixel can be understood as a pixel in the area to be compared that provides a reference basis for determining whether the target pixel is an abnormal pixel. In practical applications, the reference pixel is a pixel in the reference area corresponding to the target pixel. For example, if the reference area is a 5*5 area centered on the target pixel, the reference pixel is the other pixel in the 5*5 area except the target pixel.

[0093] After the target pixel is determined, the target pixel in the area to be compared is obtained, the disparity value to be compared of the target pixel in the area to be compared is obtained, and the average disparity to be compared corresponding to each reference pixel is obtained. The average disparity to be compared can be understood as the sum of the disparity values ​​corresponding to each reference pixel. At this point, the feature information in the area to be compared has been determined.

[0094] At the same time, the corresponding target disparity mean, target disparity variance and reference disparity mean are determined according to the target pixel in the reference comparison area. In practical applications, each pixel in the regional feature disparity image corresponds to a target disparity mean, target disparity variance and reference disparity mean. After the target pixel is determined, its corresponding target disparity mean, target disparity variance and reference disparity mean can be obtained.

[0095] In order to better compare the features and prevent the comparison errors caused by individual differences. In the method provided in the present application, the features of the target pixel in the area to be compared are not directly compared with the features in the reference comparison area. Instead, a feature fluctuation range is determined based on the features in the reference comparison area, and the feature fluctuation range is used to compare with the features in the area to be compared, thereby improving the accuracy of the comparison.

[0096] Specifically, the disparity feature value interval of the target pixel is determined according to the target disparity mean and the target disparity variance in the reference comparison area; and the reference comparison disparity value interval corresponding to the target pixel is determined according to the reference disparity mean and a preset error threshold. The disparity feature value interval can be understood as the range in which the disparity value to be compared can fluctuate, and the reference comparison disparity value interval can be understood as the range in which the disparity mean to be compared can fluctuate.

[0097] Only when the disparity value to be compared is within the disparity feature value interval, and the mean disparity value to be compared is within the reference comparison disparity value interval, can it be determined that the feature comparison of the target pixel in the area to be compared and the reference comparison area is consistent, thereby determining that the target pixel is a non-abnormal pixel. In addition, if the disparity value to be compared is not within the disparity feature value interval, or the mean disparity value to be compared is not within the reference comparison disparity value interval, it is considered that the feature comparison of the target pixel in the area to be compared and the reference comparison area is inconsistent, thereby determining that the target pixel is an abnormal pixel.

[0098] There are many methods for determining the disparity characteristic value interval of the target pixel point according to the target disparity mean and the target disparity variance. The calculation can be performed using 1sigma, 2sigma or 3sigma methods. In the method provided in this application, 1sigma is used as an example for explanation. See the following formula 3:

[0099] Formula 3

[0100] in, represents the mean target disparity, Represents the target disparity standard deviation corresponding to the target disparity variance. The disparity feature value interval corresponding to the target pixel point can be calculated by the above formula 3.

[0101] In addition, in the method provided in the present application, the reference comparison disparity value interval corresponding to the target pixel point can also be determined according to the reference disparity mean value and the preset error threshold. The preset error threshold can be a specific error value or an error percentage.

[0102] Specifically, when the preset error threshold is a specific error value, the maximum value and the minimum value of the reference comparison disparity value interval can be determined by the difference and the sum between the reference disparity mean value and the preset error threshold.

[0103] When the preset error threshold is an error percentage, the maximum and minimum values ​​of the reference comparison disparity value interval can also be determined by referring to the difference and sum of the disparity mean and the error percentage. In the method provided in the present application, the specific form of the preset error threshold is not limited.

[0104] Through the above-mentioned processing method, the terminal features of each pixel in each detection sub-region in the disparity image to be processed and the regional feature disparity image can be compared to determine the abnormal pixels, and then the abnormal proportion of the abnormal pixels is determined according to the number of abnormal pixels in each detection sub-region. The comparison value corresponding to the detection sub-region whose abnormal proportion is greater than the preset threshold is set as the first comparison value, and the comparison value corresponding to the detection sub-region whose abnormal proportion is less than or equal to the preset threshold is set as the second comparison value, thereby generating a comparison result corresponding to the disparity image to be processed.

[0105] Step 106: determining a comparison value corresponding to each detection sub-region according to the abnormal pixel points in each detection sub-region, and generating a comparison result of the parallax image to be processed according to the comparison value corresponding to each detection sub-region.

[0106] After the above steps, the abnormal pixels in each detection sub-region can be determined, and then the comparison value corresponding to each detection sub-region can be determined according to the abnormal pixels in each detection sub-region, and the comparison results of each detection sub-region can be combined to obtain the comparison result of the disparity image to be processed.

[0107] The comparison value is used to represent whether a certain detection sub-region is consistent in the disparity image to be processed and the regional feature disparity image. The comparison values ​​corresponding to each detection sub-region are combined together to generate a comparison result of the disparity image to be processed.

[0108] The comparison result can be understood as the result of whether it is consistent with the regional feature disparity image. In the method provided in the present application, each detection sub-region can be used as the comparison basis, and the comparison result can be identified in the form of an array. For example, the first comparison value can be used to represent that the detection sub-region is inconsistent with the regional feature disparity image, and the second comparison value can be used to represent that the detection sub-region is consistent with the regional feature disparity image. Furthermore, the first comparison value can be represented by 1, and the second comparison value can be represented by 0. If there are several detection sub-regions in the area to be identified, there will be several comparison values ​​in the comparison result. For example, if there are two detection sub-regions, there may be four cases for the comparison result, which are represented by four arrays [0, 0], [0, 1], [1, 0], and [1, 1] respectively. Each number in the array represents whether each detection sub-region is consistent with the regional feature disparity image. By analogy, the more detection sub-regions there are, the more numbers there are in the array.

[0109] In this embodiment, the comparison method of one detection sub-region is still used as an example for explanation, and the comparison value corresponding to each detection sub-region is determined according to the abnormal pixel points in each detection sub-region, including:

[0110] The comparison value of the target detection sub-region is determined according to the number of abnormal pixels and the total number of pixels in the target detection sub-region.

[0111] Specifically, after the comparison of each pixel in the target detection sub-region is completed, the number of abnormal pixels of the abnormal pixels is counted, and the number of abnormal pixels is compared with the total number of pixels in the target detection sub-region to determine the abnormal proportion of the number of abnormal pixels in the target detection sub-region. If the abnormal proportion exceeds a preset threshold, it means that the target detection sub-region is determined to be different from the regional feature disparity image after comparison, and there is a target object in the target detection sub-region, and the comparison value of the target detection sub-region is determined to be a first comparison value; if the abnormal proportion is less than or equal to the preset threshold, it means that the target detection sub-region is determined to be the same as the regional feature disparity image after comparison, which means that there is no target object in the target detection sub-region, and the comparison value of the target detection sub-region is determined to be a second comparison value.

[0112] After performing the above processing on each detection sub-region, a comparison value corresponding to each detection sub-region can be obtained, and the comparison values ​​corresponding to each detection sub-region are combined together to generate a comparison result of the parallax image to be processed.

[0113] Step 108: Determine whether there is a target object in the area to be identified based on the comparison result.

[0114] After determining the comparison result corresponding to the disparity image to be processed, it can be determined whether there is a target object in the area to be identified based on the comparison result. Specifically, the comparison result includes a comparison value for each detection sub-area;

[0115] Determining whether a target object exists in the area to be identified according to the comparison result includes:

[0116] When the comparison value corresponding to at least one detection sub-region is the first comparison value, determining that a target object exists in the to-be-identified region;

[0117] When the comparison value corresponding to each detection sub-region is the second comparison value, it is determined that no target object exists in the to-be-identified region.

[0118] As known from the above steps, the comparison result includes the comparison value for each detection sub-region. If the comparison value corresponding to at least one detection sub-region is the first comparison value, it can be said that the target object exists in the detection sub-region. Only when the comparison values ​​corresponding to each detection sub-region are all the second comparison values, it can be said that there is no target object in the to-be-identified region.

[0119] The detection accuracy of the method provided in the embodiment of the present application is related to the number of detection sub-regions in the region to be identified. The more the number of detection sub-regions in the region to be identified, the higher the detection accuracy.

[0120] Through the method provided in the embodiment of the present application, a disparity image is used to detect whether there is a target object in the identification area. The processing of the disparity value in the disparity image occupies less data resources, has a fast data processing speed and high efficiency, and can effectively improve the efficiency of the object detection method.

[0121] In a specific embodiment provided in the present application, the method further comprises:

[0122] Acquire a plurality of parallax images to be processed within a preset time interval to generate a sequence of parallax images to be processed;

[0123] generating a comparison result sequence according to the comparison results corresponding to the disparity image sequence to be processed and each disparity image to be processed;

[0124] The number of objects within the preset time interval is counted according to the comparison result sequence.

[0125] The above embodiment provides a detailed explanation of object detection for a parallax image to be processed. In practical applications, there are usually multiple parallax images to be processed in a time sequence. Through the method provided by the present application, multiple parallax images to be processed within a time interval can also be processed, so as to count the number of objects passing through the time interval.

[0126] Specifically, the preset time interval can be understood as a time interval in which object quantity statistics need to be performed. The image acquisition device can acquire multiple parallax images to be processed within the preset time interval, and generate a parallax image sequence to be processed according to the acquisition time.

[0127] By using the above method, the comparison results between each disparity image to be processed and the regional feature disparity image can be compared, and a comparison result sequence is generated according to the disparity image sequence to be processed and each comparison result, in which multiple comparison results are included in the comparison result sequence according to time order.

[0128] Statistics can be performed through the comparison result sequence to count the number of objects passing through the area to be identified within the preset time interval.

[0129] Specifically, the comparison result includes a comparison value for each detection sub-region;

[0130] Counting the number of objects within the preset time interval according to the comparison result sequence includes:

[0131] Matching the preset object statistical sequence with the comparison result sequence, and counting the number of successful matches;

[0132] The number of objects within the preset time interval is counted according to the number of matches.

[0133] In practical applications, each comparison structure includes comparison values ​​of each detection sub-region. For example, when there are two detection sub-regions, the comparison result sequence may be ([0, 0], [1, 0], [0, 1], [0, 0], [1, 0], [0, 1], [0, 0] as an example. The result sequence includes 7 comparison results. In practical applications, the number of comparison results is not limited and is subject to the actual application.

[0134] The comparison value in the comparison result represents whether there is an object in each detection sub-area. Taking two detection sub-areas as an example, if the comparison result sequence is not ([0, 0], [1, 0], [0, 1], [0, 0]), it can represent that an object has passed through the area to be identified in the time series. In the method provided in the present application, a preset object statistical sequence can be pre-set to indicate that the presence of the preset object statistical sequence represents that an object has passed.

[0135] The preset object statistical sequence is compared with the comparison result sequence, and the number of successful matches is counted. The number of objects that pass through the preset time interval can be counted through the number of matches.

[0136] In practical applications, the image acquisition device that obtains the disparity image can, in addition to acquiring the disparity image to be processed, also synchronously acquire the color image to be identified corresponding to the disparity image to be processed. For example, for a structured light camera, in addition to acquiring the disparity image to be processed of the area to be identified, it can also acquire a true color image, such as an RGB image. In the method provided in one embodiment of the present application, object detection can also be performed in combination with a color image. Specifically, the method further includes:

[0137] Acquire a color image to be identified corresponding to the parallax image to be processed;

[0138] Identify the color image to be identified based on each detection sub-region, and obtain a recognition result corresponding to the color image to be identified;

[0139] Accordingly, determining whether there is a target object in the to-be-identified area according to the comparison result includes:

[0140] It is determined whether there is a target object in the area to be identified according to the comparison result and the identification result.

[0141] The color image to be identified can be understood as a color image corresponding to the disparity image to be processed and for the area to be identified, which is acquired by the image acquisition device. The disparity image to be processed and the color image to be identified are aligned in time and space.

[0142] The color image to be identified also includes at least two detection sub-regions corresponding to the area to be identified. By performing object recognition on each detection sub-region, a recognition result corresponding to the color image to be identified can be obtained. In the method provided in the present application, object recognition is performed on the color image to be identified, and the color image to be identified can be processed by target detection, target tracking and other methods in traditional computer vision. In the method provided in the embodiment of the present application, the specific recognition method of the color image to be identified is not specifically limited, and is subject to actual application.

[0143] After obtaining the recognition result for the color image to be recognized, the comparison result and the recognition result can be combined to more accurately detect whether there is an object in the recognition area. The comparison result and the recognition result verify and complement each other, and a more accurate object detection result can be obtained.

[0144] In a specific implementation provided in the present application, determining whether there is a target object in the area to be identified according to the comparison result and the identification result includes:

[0145] If it is determined that the target object exists in the comparison result, or if it is determined that the target object exists in the recognition result, it is determined that the target object exists in the area to be recognized;

[0146] When it is determined that the target object does not exist in the comparison result and when it is determined that the target object does not exist in the recognition result, it is determined that the target object does not exist in the area to be recognized.

[0147] In practical applications, whether it is a comparison result or a recognition result, as long as the target object exists in one of the results, it can be determined that the target object exists in the area to be recognized. If the target object is detected in both the comparison result and the recognition result, it means that the target object does not exist in the area to be recognized.

[0148] In practical applications, the object detection method provided by the present application can be applied to the application scenario where passengers pass through a gate. In another specific implementation provided by the present application, the method further includes:

[0149] In the case where the area to be identified is a gate, the opening and closing of the gate of the gate is controlled according to the received card swiping authorization information and the determination result of whether there is a target object in the area to be identified.

[0150] When the area to be identified is a gate, the detection result of the parallax image or the detection result of the combination of the parallax image and the color image can be used to determine whether someone has passed through the gate. Combined with the card swiping authorization information received from the user who swipes the card through the card swiping area, it can be controlled whether the gate of the gate has completed the opening and closing action.

[0151] For example, the two detection sub-areas are respectively in front of and behind the gate of the gate machine. When the card swipe authorization information is received, the gate of the gate machine can be controlled to open. When the corresponding comparison result of a certain disparity image to be processed is [0, 1], and the card swipe authorization information is received for more than the preset time threshold, the gate of the gate machine can be controlled to close.

[0152] In actual applications, when controlling the opening and closing of the gate, the number of people passing through within the preset time interval is counted, and by comparing it with the number of card swiping authorization information obtained at the current moment, it is determined whether the gate should be closed and a closing control signal is issued.

[0153] At the same time, the method provided by the present application can also count the number of passengers passing through the gate. In addition, the result of combining the parallax image and the color image can also be used as a basis for judging the gate's passage safety zone. If there is a target detection frame of a person passing through the gate's passage safety zone or the comparison value corresponding to a certain inspection sub-area is the first comparison value, it means that there are passengers in the gate's passage safety zone and the gate should not be closed.

[0154] Through the method provided in the embodiment of the present application, a disparity image is used to detect whether there is a target object in the identification area. The processing of the disparity value in the disparity image occupies less data resources, has a fast data processing speed and high efficiency, and can effectively improve the efficiency of the object detection method.

[0155] In addition, the color image corresponding to the disparity image can be combined to determine whether there is a target object in the area to be identified, and the comparison result and the recognition result can be combined to perform a more accurate detection of whether there is an object in the area to be identified. The comparison result and the recognition result verify and complement each other to obtain a more accurate object detection result.

[0156] After improving the accuracy of object detection, the traffic conditions of passengers can be better judged for the scenes of gate opening and closing, reducing misjudgments and missed judgments, and reducing the harm caused to passengers by gate closures. Through the object detection of this method, the number of people passing through the gate can be counted. Combined with the number of authorization information of the gate, the gate can be closed accurately, and when closing the gate, it will not hit passengers who have not yet passed. Through reliable traffic judgment and accurate door opening and closing control, traffic congestion and accidents caused by improper gate operation can be effectively avoided. Improve the smoothness of traffic in rail transit stations.

[0157] The following combination Figure 3 , taking the application of the object detection method provided by this application in the gate switch control scenario as an example, the object detection method is further explained. Figure 3A processing flow chart of an object detection method applied to a gate switch control scene provided by an embodiment of the present application is shown, which specifically includes the following steps:

[0158] Step 302: Acquire a plurality of parallax images to be processed and a plurality of color images to be identified corresponding to the plurality of parallax images to be processed for the gate area within a preset time interval, wherein the preset time interval is determined by the current time point.

[0159] Step 304: Acquire a regional feature parallax image for the gate area, wherein the gate area includes at least two detection sub-areas.

[0160] Step 306: comparing each to-be-processed disparity image with the regional characteristic disparity image based on the disparity value characteristic information of the pixel points in each detection sub-region, and determining abnormal pixel points in each detection sub-region.

[0161] Step 308: determining a comparison value corresponding to each detection sub-region according to abnormal pixels in each detection sub-region, and generating a comparison result of the parallax image to be processed according to the comparison value corresponding to each detection sub-region.

[0162] Step 310: Identify each color image to be identified based on each detection sub-region, and obtain a recognition result corresponding to each color image to be identified.

[0163] Step 312: Count the number of people who pass through within a preset time interval according to each comparison result and the recognition result corresponding to each comparison result.

[0164] Step 314: Obtain the card swipe quantity of the card swipe authorization information corresponding to the gate area.

[0165] Step 316: When it is determined that the gate is closed based on the number of cards swiped and the number of people passing through, determine whether there are any passengers in the gate safety zone based on the parallax image to be processed and the color image to be recognized corresponding to the current time point.

[0166] See also Figure 4 , Figure 4 FIG. 1 shows a schematic diagram of a passenger passing through a gate outside a safety zone provided by an embodiment of the present application. Figure 4 In the case shown, it can be determined that the passenger has passed through the gate and is outside the gate's safety zone.

[0167] Step 318: When it is determined that there are no passengers in the gate safety zone, the gate of the gate is controlled to close.

[0168] Through the method provided in the embodiment of the present application, for the scene of gate opening and closing, the passenger's passage situation can be better judged, misjudgment and missed judgment can be reduced, and the damage caused to passengers by the gate closing by mistake can be reduced. Through the object detection of this method, the number of people passing through the gate can be counted. Combined with the number of authorization information of the gate, the gate can be closed accurately, and when closing the gate, it will not hit passengers who have not passed. Through reliable passage judgment and accurate door opening and closing control, traffic congestion and accidents caused by improper gate operation can be effectively avoided. Improve the smoothness of traffic in rail transit stations.

[0169] Corresponding to the above method embodiment, the present application also provides an object detection device embodiment, Figure 5 FIG. 1 is a schematic diagram showing the structure of an object detection device provided by an embodiment of the present application. Figure 5 As shown, the device comprises:

[0170] An acquisition module 502 is configured to acquire a to-be-processed disparity image and a regional feature disparity image for a region to be identified, wherein the region to be identified includes at least two detection sub-regions;

[0171] A determination module 504 is configured to compare the to-be-processed disparity image with the regional feature disparity image based on the disparity value feature information of the pixel points in each detection sub-region, and determine the abnormal pixel points in each detection sub-region;

[0172] A generating module 506 is configured to determine a comparison value corresponding to each detection sub-region according to abnormal pixels in each detection sub-region, and generate a comparison result of the parallax image to be processed according to the comparison value corresponding to each detection sub-region;

[0173] The detection module 508 is configured to determine whether there is a target object in the area to be identified according to the comparison result.

[0174] Optionally, the comparison result includes a comparison value for each detection sub-region;

[0175] The detection module 508 is further configured to:

[0176] When the comparison value corresponding to at least one detection sub-region is the first comparison value, determining that a target object exists in the to-be-identified region;

[0177] When the comparison value corresponding to each detection sub-region is the second comparison value, it is determined that no target object exists in the to-be-identified region.

[0178] Optionally, the device further includes a statistical module configured to:

[0179] Acquire a plurality of parallax images to be processed within a preset time interval to generate a sequence of parallax images to be processed;

[0180] generating a comparison result sequence according to the comparison results corresponding to the disparity image sequence to be processed and each disparity image to be processed;

[0181] The number of objects within the preset time interval is counted according to the comparison result sequence.

[0182] Optionally, the comparison result includes a comparison value for each detection sub-region;

[0183] The statistics module is further configured as follows:

[0184] Matching the preset object statistical sequence with the comparison result sequence, and counting the number of successful matches;

[0185] The number of objects within the preset time interval is counted according to the number of matches.

[0186] Optionally, the device further includes an identification module configured to:

[0187] Acquire a color image to be identified corresponding to the parallax image to be processed;

[0188] Identify the color image to be identified based on each detection sub-region, and obtain a recognition result corresponding to the color image to be identified;

[0189] Accordingly, the detection module 508 is further configured to:

[0190] It is determined whether there is a target object in the area to be identified according to the comparison result and the identification result.

[0191] Optionally, the detection module 508 is further configured to:

[0192] If it is determined that the target object exists in the comparison result, or if it is determined that the target object exists in the recognition result, it is determined that the target object exists in the area to be recognized;

[0193] When it is determined that the target object does not exist in the comparison result and when it is determined that the target object does not exist in the recognition result, it is determined that the target object does not exist in the area to be recognized.

[0194] Optionally, the acquisition module 502 is further configured to:

[0195] Acquire a plurality of static area disparity images for the area to be identified, wherein the static area disparity images are images in which no target object exists in the area to be identified;

[0196] The pixel feature information of two detection sub-regions in each of the multiple static regional disparity images is fused to generate a regional feature disparity image.

[0197] Optionally, the acquisition module 502 is further configured to:

[0198] Determine a target pixel point in the target detection sub-region and a reference pixel point corresponding to the target pixel point;

[0199] The target disparity value of the target pixel point and the reference disparity value of each reference pixel point in each static area disparity image;

[0200] Determine pixel feature information of the target pixel point according to each target disparity value and each reference disparity value;

[0201] A regional feature disparity image is generated according to the pixel feature information of each target pixel in each detection sub-region.

[0202] Optionally, the acquisition module 502 is further configured to:

[0203] Determining a target disparity mean and a target disparity variance corresponding to the target pixel point according to the target disparity value of the target pixel point in each static area disparity image;

[0204] According to the disparity values ​​of the reference pixels in the disparity images of the static regions, the initial reference disparity mean values ​​corresponding to the reference pixels are calculated, and according to the initial reference disparity mean values, the reference disparity mean values ​​corresponding to the target pixels are calculated;

[0205] The pixel feature information of the target pixel point is determined according to the target disparity mean, the target disparity variance and the reference disparity mean.

[0206] Optionally, the determining module 504 is further configured to:

[0207] Determine, according to the target detection sub-region, a region to be compared in the to-be-processed disparity image and a reference comparison region in the regional feature disparity image;

[0208] Determine abnormal pixels according to the disparity value feature information of each pixel in the to-be-compared area and the reference comparison disparity feature of each pixel in the reference comparison area;

[0209] Accordingly, the generating module 506 is further configured to:

[0210] The comparison value of the target detection sub-region is determined according to the number of abnormal pixels and the total number of pixels in the target detection sub-region.

[0211] Optionally, the determining module 504 is further configured to:

[0212] Determine a target pixel and at least one reference pixel corresponding to the target pixel, wherein the target pixel is any pixel in the area to be compared, and each reference pixel is a pixel in the reference area corresponding to the target pixel;

[0213] Obtaining a disparity value to be compared of the target pixel in the area to be compared and an average disparity value to be compared between each reference pixel;

[0214] Obtaining a target disparity mean, a target disparity variance, and a reference disparity mean of a target pixel point in the reference comparison area;

[0215] Determine the disparity feature value interval of the target pixel point according to the target disparity mean and the target disparity variance, and determine the reference comparison disparity value interval corresponding to the target pixel point according to the reference disparity mean and a preset error threshold;

[0216] When the to-be-compared disparity value is not within the disparity characteristic value interval, or the to-be-compared disparity mean value is not within the reference comparison disparity value interval, it is determined that the target pixel is an abnormal pixel.

[0217] Optionally, the device further comprises:

[0218] The control module is configured to control the opening and closing of the gate of the gate according to the received card swiping authorization information and the determination result of whether there is a target object in the area to be identified when the area to be identified is a gate.

[0219] Through the device provided in the embodiment of the present application, a disparity image method is used to detect whether there is a target object in the identification area. The processing of the disparity value in the disparity image occupies less data resources, has a fast data processing speed and high efficiency, and can effectively improve the efficiency of the object detection method.

[0220] In addition, the color image corresponding to the disparity image can be combined to determine whether there is a target object in the area to be identified, and the comparison result and the recognition result can be combined to perform a more accurate detection of whether there is an object in the area to be identified. The comparison result and the recognition result verify and complement each other to obtain a more accurate object detection result.

[0221] After improving the accuracy of object detection, the traffic conditions of passengers can be better judged for the scenes of gate opening and closing, reducing misjudgments and missed judgments, and reducing the harm caused to passengers by gate closures. Through the object detection of this method, the number of people passing through the gate can be counted. Combined with the number of authorization information of the gate, the gate can be closed accurately, and when closing the gate, it will not hit passengers who have not yet passed. Through reliable traffic judgment and accurate door opening and closing control, traffic congestion and accidents caused by improper gate operation can be effectively avoided. Improve the smoothness of traffic in rail transit stations.

[0222] The above is a schematic scheme of an object detection device of this embodiment. It should be noted that the technical scheme of the object detection device and the technical scheme of the object detection method described above are of the same concept, and the details not described in detail in the technical scheme of the object detection device can be referred to the description of the technical scheme of the object detection method described above.

[0223] Figure 6 The block diagram of a computing device 600 according to an embodiment of the present application is shown. The components of the computing device 600 include but are not limited to a memory 610 and a processor 620. The processor 620 is connected to the memory 610 via a bus 630, and the database 650 is used to store data.

[0224] The computing device 600 also includes an access device 640 that enables the computing device 600 to communicate via one or more networks 660. Examples of these networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 640 may include one or more of any type of network interface (e.g., a network interface card (NIC)) that is wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a world-wide interoperability for microwave access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, a near field communication (NFC) interface, and the like.

[0225] In one embodiment of the present application, the above components of the computing device 600 and Figure 6 Other components not shown in the figure may also be connected to each other, for example, via a bus. It should be understood that Figure 6 The computing device structure block diagram shown is only for the purpose of illustration, and is not intended to limit the scope of the present application. Those skilled in the art may add or replace other components as needed.

[0226] The computing device 600 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, a personal digital assistant, a laptop computer, a notebook computer, a netbook, etc.), a mobile phone (e.g., a smart phone), a wearable computing device (e.g., a smart watch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or a personal computer (PC). The computing device 600 may also be a mobile or stationary server.

[0227] The processor 620 is used to execute the following computer program / instructions, which implement the steps of the above-mentioned object detection method when executed by the processor.

[0228] The above is a schematic scheme of a computing device of this embodiment. It should be noted that the technical scheme of the computing device and the technical scheme of the above object detection method belong to the same concept, and the details not described in detail in the technical scheme of the computing device can be referred to the description of the technical scheme of the above object detection method.

[0229] An embodiment of the present specification further provides a computer-readable storage medium storing a computer program / instruction, which implements the steps of the above-mentioned object detection method when executed by a processor.

[0230] The above is a schematic scheme of a computer-readable storage medium of this embodiment. It should be noted that the technical scheme of the storage medium and the technical scheme of the object detection method described above are of the same concept, and the details not described in detail in the technical scheme of the storage medium can be referred to the description of the technical scheme of the object detection method described above.

[0231] An embodiment of the present specification also provides a computer program product, including a computer program / instruction, which implements the steps of the above-mentioned object detection method when executed by a processor.

[0232] The above is a schematic solution of a computer program product of this embodiment. It should be noted that the technical solution of the computer program product and the technical solution of the object detection method described above are of the same concept, and the details not described in detail in the technical solution of the computer program product can be found in the description of the technical solution of the object detection method described above.

[0233] The above describes specific embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0234] The computer instructions include computer program codes, which may be in source code form, object code form, executable files or some intermediate forms, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of patent practice. For example, in some regions, according to patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0235] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described action sequence, because according to the present application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0236] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0237] The preferred embodiments of the present application disclosed above are only used to help explain the present application. The optional embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of the present application. The present application selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present application, so that those skilled in the art can understand and use the present application well. The present application is only limited by the claims and their full scope and equivalents.

Claims

1. An object detection method, characterized in that: include: Acquire a to-be-processed disparity image and a regional feature disparity image for a region to be identified, wherein the region to be identified includes at least two detection sub-regions; Based on the disparity value feature information of the pixel points in each detection sub-region, the disparity image to be processed is compared with the regional feature disparity image to determine the abnormal pixel points in each detection sub-region; Determining a comparison value corresponding to each detection sub-region according to abnormal pixel points in each detection sub-region, and generating a comparison result of the parallax image to be processed according to the comparison value corresponding to each detection sub-region; Determine whether there is a target object in the area to be identified according to the comparison result.

2. The method according to claim 1, characterized in that The comparison result includes a comparison value for each detection sub-region; Determining whether a target object exists in the area to be identified according to the comparison result includes: When the comparison value corresponding to at least one detection sub-region is the first comparison value, determining that a target object exists in the to-be-identified region; When the comparison value corresponding to each detection sub-region is the second comparison value, it is determined that no target object exists in the to-be-identified region.

3. The method according to claim 1, characterized in that The method further comprises: Acquire a plurality of parallax images to be processed within a preset time interval to generate a sequence of parallax images to be processed; generating a comparison result sequence according to the comparison results corresponding to the disparity image sequence to be processed and each disparity image to be processed; The number of objects within the preset time interval is counted according to the comparison result sequence.

4. The method according to claim 3, characterized in that The comparison result includes a comparison value for each detection sub-region; Counting the number of objects within the preset time interval according to the comparison result sequence includes: Matching the preset object statistical sequence with the comparison result sequence, and counting the number of successful matches; The number of objects within the preset time interval is counted according to the number of matches.

5. The method according to claim 1, characterized in that The method further comprises: Acquire a color image to be identified corresponding to the parallax image to be processed; Identify the color image to be identified based on each detection sub-region, and obtain a recognition result corresponding to the color image to be identified; Accordingly, determining whether there is a target object in the to-be-identified area according to the comparison result includes: It is determined whether there is a target object in the area to be identified according to the comparison result and the identification result.

6. The method according to claim 5, characterized in that Determining whether a target object exists in the area to be identified according to the comparison result and the identification result includes: If it is determined that the target object exists in the comparison result, or if it is determined that the target object exists in the recognition result, it is determined that the target object exists in the area to be recognized; When it is determined that the target object does not exist in the comparison result and when it is determined that the target object does not exist in the recognition result, it is determined that the target object does not exist in the area to be recognized.

7. The method according to claim 1, characterized in that Acquiring a regional feature disparity image for the area to be identified, including: Acquire a plurality of static area disparity images for the area to be identified, wherein the static area disparity images are images in which no target object exists in the area to be identified; The pixel feature information of two detection sub-regions in each of the multiple static regional disparity images is fused to generate a regional feature disparity image.

8. The method according to claim 7, characterized in that The pixel feature information of two detection sub-regions in each of the multiple static regional disparity images is fused to generate a regional feature disparity image, including: Determine a target pixel point in the target detection sub-region and a reference pixel point corresponding to the target pixel point; The target disparity value of the target pixel point and the reference disparity value of each reference pixel point in each static area disparity image; Determine pixel feature information of the target pixel point according to each target disparity value and each reference disparity value; A regional feature disparity image is generated according to the pixel feature information of each target pixel in each detection sub-region.

9. The method according to claim 8, characterized in that Determining pixel feature information of the target pixel point according to each target disparity value and each reference disparity value includes: Determining a target disparity mean and a target disparity variance corresponding to the target pixel point according to the target disparity value of the target pixel point in each static area disparity image; According to the disparity values ​​of the reference pixels in the disparity images of the static regions, the initial reference disparity mean values ​​corresponding to the reference pixels are calculated, and according to the initial reference disparity mean values, the reference disparity mean values ​​corresponding to the target pixels are calculated; The pixel feature information of the target pixel point is determined according to the target disparity mean, the target disparity variance and the reference disparity mean.

10. The method according to claim 1, characterized in that The method of comparing the to-be-processed disparity image with the regional feature disparity image based on the disparity value feature information of the pixel points in each detection sub-region to determine the abnormal pixel points in each detection sub-region includes: Determine, according to the target detection sub-region, a region to be compared in the to-be-processed disparity image and a reference comparison region in the regional feature disparity image; Determine abnormal pixels according to the disparity value feature information of each pixel in the to-be-compared area and the reference comparison disparity feature of each pixel in the reference comparison area; Accordingly, the comparison value corresponding to each detection sub-region is determined according to the abnormal pixel points in each detection sub-region, including: The comparison value of the target detection sub-region is determined according to the number of abnormal pixels and the total number of pixels in the target detection sub-region.

11. The method according to claim 10, characterized in that Determining abnormal pixels according to the disparity value feature information of each pixel in the to-be-compared area and the reference comparison disparity feature of each pixel in the reference comparison area includes: Determine a target pixel and at least one reference pixel corresponding to the target pixel, wherein the target pixel is any pixel in the area to be compared, and each reference pixel is a pixel in the reference area corresponding to the target pixel; Obtaining a disparity value to be compared of the target pixel in the area to be compared and an average disparity value to be compared between each reference pixel; Obtaining a target disparity mean, a target disparity variance, and a reference disparity mean of a target pixel point in the reference comparison area; Determine the disparity feature value interval of the target pixel point according to the target disparity mean and the target disparity variance, and determine the reference comparison disparity value interval corresponding to the target pixel point according to the reference disparity mean and a preset error threshold; When the to-be-compared disparity value is not within the disparity characteristic value interval, or the to-be-compared disparity mean value is not within the reference comparison disparity value interval, it is determined that the target pixel is an abnormal pixel.

12. The method according to any one of claims 1 to 11, characterized in that: The method further comprises: In the case where the area to be identified is a gate, the opening and closing of the gate of the gate is controlled according to the received card swiping authorization information and the determination result of whether there is a target object in the area to be identified.

13. An object detection device, characterized in that: include: An acquisition module is configured to acquire a to-be-processed disparity image and a regional feature disparity image for a region to be identified, wherein the region to be identified includes at least two detection sub-regions; A determination module is configured to compare the to-be-processed disparity image with the regional characteristic disparity image based on the disparity value characteristic information of the pixel points in each detection sub-region, and determine the abnormal pixel points in each detection sub-region; A generating module, configured to determine a comparison value corresponding to each detection sub-region according to abnormal pixel points in each detection sub-region, and generate a comparison result of the parallax image to be processed according to the comparison value corresponding to each detection sub-region; The detection module is configured to determine whether there is a target object in the area to be identified according to the comparison result.

14. A computing device, characterized in that include: Memory and processor; The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions. When the computer programs / instructions are executed by the processor, the steps of the method described in any one of claims 1 to 12 are implemented.

15. A computer-readable storage medium storing a computer program / instruction, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.

16. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.

Images