CN114842213A - An obstacle contour detection method, device, terminal device and storage medium - Google Patents

Abstract

The embodiment of the invention discloses a method, a device, a terminal device and a storage medium for detecting an obstacle contour, wherein the method for detecting the obstacle contour comprises the steps of acquiring an actual image of an actual scene through a monocular camera, obtaining an actual aerial view through perspective transformation, calculating actual characteristic values of all pixel points in the actual aerial view, obtaining a binary image based on the aerial characteristic values and the actual characteristic values of all the pixel points of the aerial view, denoising abnormal points in the binary image to obtain a first image, and detecting the contour of an obstacle in the actual scene through the contour of the first image. The detection method can accurately identify the size and the appearance characteristics of the barrier, can remove the influence of illumination on image characteristic identification, can be applied to most barrier characteristic identification, and avoids using a large amount of algorithm training and testing, thereby reducing the production cost and the labor cost.

Description

An obstacle contour detection method, device, terminal device and storage medium

technical field

The present invention relates to the field of obstacle detection, and in particular, to a method, device, terminal device and storage medium for obstacle contour detection.

Background technique

In the prior art, a binocular camera is generally used for obstacle detection, and an AI (Artificial Intelligence, artificial intelligence) recognition algorithm is used for calculation. However, the binocular camera is complex in structure, high in production cost, and requires high computing power. To use the AI recognition algorithm, it needs to be pre-trained, and requires high processor storage and computing power. When using a binocular camera, due to the need for Adapting to different lighting and different obstacles has high requirements for actual use, and requires a large number of scenarios for algorithm training and testing, which is more complicated to use.

SUMMARY OF THE INVENTION

In view of this, the present application provides an obstacle contour detection method, device, terminal device and storage medium.

In a first aspect, the present application proposes a method for detecting the contour of an obstacle with a monocular camera, the method comprising:

Obtain an actual image of the actual scene through a monocular camera, and perform perspective transformation on the actual image to obtain an actual bird's-eye view;

Calculate the actual feature value of each pixel in the actual bird's-eye view;

Perform outlier identification processing based on the open space feature value of each pixel point of the open space bird's-eye view and the actual characteristic value, and obtain a binary image containing outliers;

Denoising the abnormal points in the binary image to obtain a first image;

Perform contour detection on the first image to determine the contour of the obstacle in the actual scene.

In some embodiments, the performing contour detection on the first image to determine the contour of the obstacle in the actual scene includes:

extracting at least one contour in the first image based on a preset algorithm;

Calculate the center point of each of the contours and the side length and area of the smallest oblique rectangle;

If the center point of the current contour is not at the boundary of the first image, detecting whether the side length is less than a preset side length threshold, and whether the area is less than a preset area threshold;

If the side length is less than the side length threshold and the area is less than the area threshold, then the current contour is determined to be an interference contour and discarded, otherwise the current contour is retained;

If the center point of the current contour is at the boundary of the first image, retaining the current contour;

Take the remaining contour as the contour of the obstacle.

In some embodiments, before acquiring the actual image of the actual scene through the monocular camera, it includes:

Obtain an open space image of the open space scene through a monocular camera, and measure the actual size of the open space scene to determine the aspect ratio of the open space scene;

Determine the coordinates of the open space based on the aspect ratio and the resolution of the open space aerial view of the open space scene to be generated;

Setting a corner point at the boundary of the open space scene, and determining the corner point coordinates corresponding to the corner point in the open space image;

Calculate the parameters of the perspective transformation matrix based on the corner coordinates and the open space coordinates to determine the perspective transformation matrix;

Perspective transformation is performed on the open space image through the perspective transformation matrix to obtain an open space bird's-eye view of the open space scene.

In some embodiments, the eigenvalues of each pixel in the corresponding bird's-eye view are calculated in the following manner:

Perform grayscale processing on the corresponding bird's-eye view to obtain a grayscale image;

Calculate the corresponding center brightness and ambient brightness based on the respective pixel points of the grayscale image as the center;

According to the ratio of the central brightness of each pixel to the ambient brightness, the feature value of the corresponding pixel is taken.

In some embodiments, the process of identifying outliers based on the open space feature values of each pixel point of the open space bird's eye view and the actual feature value includes:

Comparing the open space feature value and the actual feature value of the pixel point at the same position to obtain a corresponding ratio;

If the ratio is less than or equal to a preset ratio threshold, the pixel point is determined to be the abnormal point; otherwise, the pixel point is determined to be a normal point.

In some embodiments, the performing denoising processing on the abnormal points in the binary image to obtain the first image includes:

Dilation processing of the first pixel size is performed on the binary image to obtain a second image;

Erosion processing of the second pixel size is performed on the second image to obtain a third image;

Dilation processing of a third pixel size is performed on the third image to obtain a fourth image;

Erosion processing of a fourth pixel size is performed on the fourth image to obtain a first image;

The relationship between the first pixel size, the second pixel size, the third pixel size and the fourth pixel size is as follows: the sum of the first pixel size and the third pixel size is equal to the The sum of the second pixel size and the fourth pixel size, the first pixel size is smaller than the second pixel size is smaller than the third pixel size.

In some embodiments, the etching process includes:

The center point of the preset structural element is sequentially overlapped with each abnormal point in the binary image, and it is judged whether all the pixel points in the structural element are abnormal points;

If all the pixel points are the abnormal points, keep the pixel points corresponding to the center point;

Otherwise, modify the pixel point corresponding to the center point to a normal point.

In a second aspect, an embodiment of the present application further provides a device for judging an obstacle contour, including:

The bird's-eye view acquisition module obtains the actual image of the actual scene through the monocular camera, and performs perspective transformation on the actual image to obtain the actual bird's-eye view;

an eigenvalue calculation module, which calculates the actual eigenvalues of each pixel in the actual bird's-eye view;

The outlier identification module performs outlier identification processing based on the open space characteristic value of each pixel point of the open space bird's eye view and the actual characteristic value, and obtains a binary image containing outliers;

a denoising processing module, which performs denoising processing on the abnormal points in the binary image to obtain a first image;

The contour determination module performs contour detection on the first image to determine the contour of the obstacle in the actual scene.

In a third aspect, an embodiment of the present application further provides a terminal device, including a memory and a processor, wherein the memory stores a computer program, and the computer program executes any of the above obstacle contour detection when running on the processor method.

In a fourth aspect, an embodiment of the present application further provides a readable storage medium, which stores a computer program, and the computer program executes any one of the above-mentioned obstacle contour detection methods when running on a processor.

The embodiments of the present application have the following beneficial effects:

Embodiments of the present application provide a method, device, terminal device, and storage medium for detecting an obstacle contour, obtaining an actual image of an actual scene through a monocular camera, performing perspective transformation on the actual image to obtain an actual bird's-eye view, and calculating the actual bird's-eye view. The actual eigenvalues of each pixel in the open-air bird's-eye view are used to identify and process outliers based on the open-space eigenvalues and actual eigenvalues of each pixel in the aerial view of the open space, and a binary image containing outliers is obtained. The first image is obtained by noise processing, and contour detection is performed on the first image to determine the contour of the obstacle in the actual scene. On the one hand, the detection method of the present application can more accurately determine the size and shape features of the identified obstacles, and on the other hand, the denoising process of abnormal points can remove the influence of illumination on image feature recognition, and can be applied to most Obstacle feature recognition avoids the use of a large number of algorithm training and testing, which can reduce production costs and labor costs.

Description of drawings

In order to illustrate the technical solutions of the present invention more clearly, the accompanying drawings required in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be It is regarded as the limitation of the protection scope of the present invention. In the various figures, similar components are numbered similarly.

FIG. 1 shows a schematic flowchart of an obstacle contour detection method proposed by an embodiment of the present application;

2 shows a schematic flowchart of generating a bird's-eye view of an open space in an obstacle contour detection method proposed by an embodiment of the present application;

FIG. 3 shows a schematic flowchart of calculating feature values in an obstacle contour detection method proposed by an embodiment of the present application;

FIG. 4 shows a schematic flowchart of abnormal point identification processing in an obstacle contour detection method proposed by an embodiment of the present application;

FIG. 5 shows a schematic flowchart of corrosion processing in an obstacle contour detection method proposed by an embodiment of the present application;

FIG. 6 shows a schematic flowchart of an obstacle contour determination in an obstacle contour detection method proposed by an embodiment of the present application;

FIG. 7 shows a schematic diagram of an obstacle in an obstacle contour detection method proposed by an embodiment of the present application;

FIG. 8 shows a schematic structural diagram of an obstacle contour determination device provided by an embodiment of the present application.

Description of main component symbols:

10-obstacle outline judgment device; 11-bird's-eye view acquisition module; 12-eigenvalue calculation module; 13-abnormal point identification module; 14-denoising processing module; 15-contour determination module.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments.

The components of the embodiments of the invention generally described and illustrated in the drawings herein may be arranged and designed in a variety of different configurations. Thus, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but is merely representative of selected embodiments of the invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the present invention.

Hereinafter, the terms "comprising", "having" and their cognates, which may be used in various embodiments of the present invention, are only intended to denote particular features, numbers, steps, operations, elements, components, or combinations of the foregoing, and should not be construed as first excluding the presence of or adding one or more other features, numbers, steps, operations, elements, components or combinations of the foregoing or the possibility of a combination of the foregoing.

Furthermore, the terms "first", "second", "third", etc. are only used to differentiate the description and should not be construed as indicating or implying relative importance.

Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which various embodiments of this invention belong. The terms (such as those defined in commonly used dictionaries) will be interpreted as having the same meaning as the contextual meaning in the relevant technical field and will not be interpreted as having an idealized or overly formal meaning, unless explicitly defined in the various embodiments of the present invention.

Example 1

An embodiment of the present application, as shown in FIG. 1 , provides a method for detecting the contour of an obstacle with a monocular camera, including steps S110-S150:

S110: Acquire an actual image of the actual scene through a monocular camera, and perform perspective transformation on the actual image to obtain an actual bird's-eye view.

In this embodiment, the object is re-projected to a new imaging plane by transforming according to the imaging projection law of the object. In other words, the perspective transformation is performed on the photo obtained from the perspective of the monocular camera to obtain the top plan view of the open space scene, that is, the bird's-eye view, so as to It is used to detect the target features with a unified benchmark, so as to realize the detection of obstacles.

Wherein, as shown in FIG. 2 , steps S210 to S250 are further included before step S100:

S210: Acquire an open space image of the open space scene by using a monocular camera, and measure the actual size of the open space scene to determine the aspect ratio of the open space scene.

In this embodiment, the open space image of the open space scene is obtained through a monocular camera, the actual size of the open space scene is measured, and the length-width ratio of the actual site is determined by measuring the actual size of the open space scene.

S220: Determine the coordinates of the open space based on the aspect ratio and the resolution of the open space bird's-eye view of the open space scene to be generated.

According to the aspect ratio of the open space scene and the maximum resolution of the open space aerial view to be generated, the length and width of the open space scene at the maximum resolution can be calculated, so that the open space coordinates corresponding to the open space scene can be determined.

Exemplarily, if the actual length-to-width ratio of the open space scene is m:n by measurement, and the length L and the width of the bird's-eye view of the open space to be generated at the maximum resolution are obtained as L and the width as W, the length L of the open space at the maximum resolution can be calculated. ' and width W'. The open space coordinates obtained after the perspective transformation of the open space scene can be determined by the actual aspect ratio m:n and the length L and width W of the open space image. The above open space coordinates include: (0, 0), (0, L'), (W' , 0), (W', L'). Wherein, when W*m/n<L, the length L' and width W' are obtained as L'=W*m/n, W'=W; otherwise, the length L' and width W' are obtained as L respectively '=L, W'=L*n/m.

In this embodiment, the cost of acquiring images through the monocular camera is lower, and in actual use, algorithm training and testing in a large number of scenarios are not required, which can reduce the workload of the staff.

S230: Set a corner point at the boundary of the open space scene, and determine the corner point coordinates corresponding to the corner point in the open space image.

It can be understood that the corner points are extreme points, that is, points with particularly prominent properties in some aspects of the image. In this embodiment, multiple corner points may be set at the boundary of the open space scene, or the intersection of two lines may be used as the corner point, or the point located on two adjacent things with different main directions may be used as the corner point. corner. The coordinates of the corresponding corner points in the open-air image can be determined through a preset detection algorithm, wherein the above detection algorithms include but are not limited to detection algorithms such as Harris corner algorithm and Shi-Tomas corner algorithm.

Exemplarily, at least four corner points may be set at the boundary of the open space scene. For example, if a pure black block with a size of 2cm*2cm is placed at the boundary, the pure black block may be a corner point in the image. The coordinates of the four corner points are automatically identified by the Harris corner point algorithm as (x1, y1), (x2, y1), (x1, y2), (x2, y2), where x represents the length direction and y represents the width direction.

S240: Calculate parameters of a perspective transformation matrix based on the corner coordinates and the open space coordinates to determine the perspective transformation matrix.

Perspective transformation is to project a two-dimensional picture onto a three-dimensional view plane, and then convert it to two-dimensional coordinates, also known as projection mapping.

Specifically, the perspective transformation matrix is as follows:

Among them, X, Y and Z represent the three-dimensional coordinates after perspective transformation, and x and y represent the two-dimensional coordinates before perspective transformation. Through the above perspective transformation matrix formula, we can get:

Because (X, Y, Z) are three-dimensional coordinates, it is necessary to convert the obtained three-dimensional coordinates into two-dimensional coordinates (x', y', 1), so as to obtain the following formula:

Therefore, x' and y' are the final calculation results of the two-dimensional perspective transformation, where c3=1. Take the above empty field coordinates and corner coordinates as the coordinate points before and after perspective transformation into the above formula, namely (0, 0), (0, L'), (W', 0), (W', L') and ( x1, y1), (x2, y1), (x1, y2), (x2, y2) are brought into the above formula as 4 groups (x, y), (x', y'), and the parameters of the perspective transformation matrix are calculated. a1, a2, a3, b1, b2, b3 and c1.

S250: Perform perspective transformation on the open space image by using the perspective transformation matrix to obtain an open space bird's-eye view of the open space scene.

Through the above perspective transformation formula, any point (x, y) of the image obtained by the monocular camera can be brought in, and the coordinate point (x', y') after perspective transformation can be calculated. In other words, the collected image of the open space is subjected to perspective transformation through the above perspective transformation matrix to obtain a bird's-eye view of the open space of the open space scene.

By transforming according to the imaging projection law of the object, the object is re-projected to a new imaging plane. In other words, the perspective transformation of the photo obtained from the perspective of the monocular camera is performed to obtain the top plan view of the open space scene, that is, the bird's-eye view, so as to use a unified reference to compare The target feature is detected, that is, the obstacle is detected.

For the above step S110, exemplarily, after obtaining the open space image of the open space scene, keep the installation position of the monocular camera unchanged, obtain the actual image of the actual scene through the monocular camera, and use the above perspective transformation matrix to transform the actual image. Perspective transformation processing is performed to obtain the top plan view of the actual scene, that is, the actual bird's-eye view.

S120: Calculate the actual feature value of each pixel in the actual bird's-eye view.

It can be understood that, after obtaining the actual bird's-eye view, the actual feature values of each pixel in the actual bird's-eye view corresponding to the actual scene are determined by calculation. The above-mentioned actual feature values and the open space characteristic values of each pixel of the open space bird's-eye view corresponding to the open space scene can be calculated by the same calculation method. Among them, as shown in Figure 3, the eigenvalues of each pixel in the corresponding bird's-eye view of the actual scene and the open space scene can be calculated by the following sub-steps:

Sub-step S121: Perform grayscale processing on the corresponding bird's-eye view to obtain a grayscale image.

It can be understood that after the perspective transformation, the bird's-eye view corresponding to the open space scene and the actual scene is obtained, and the bird's-eye view is converted into a grayscale image through grayscale processing. The value ranges from 0 to 255, and the grayscale processing methods for the corresponding bird's-eye view include: component method, maximum value method, average method and weighted average method.

Sub-step S122: Calculate the corresponding center brightness and ambient brightness based on the pixels of the grayscale image as the center.

After obtaining the grayscale image, take each pixel in the grayscale image as the center, take the pixels in the first range around the pixel as the core area of the pixel, and calculate the average brightness of the core area of each pixel is the central brightness of the pixel, that is, the pixel is taken as the core, and the average value of the gray values of all the pixels in the core area is taken. Take the surrounding second range of pixels as the environmental area of the pixel, and calculate the average brightness of the environmental area of each pixel as the environmental brightness of the pixel, that is, calculate all the gray values in the environmental area around each pixel average of. Wherein, the pixel points of the first range are smaller than the pixel points of the second range.

Exemplarily, when the pixel point (30, 30) is the core point and the range value of the core area is 3*3, then the central brightness of the pixel point is the pixel point (29, 29), (29, 30), ( 29, 31), (30, 29), (30, 30), (30, 31), (31, 29), (31, 30), (31, 31) a total of 9 points of image gray value average value. When the range value of the environment area is 9*9, the average value of the gray values of a total of 81 pixels in the surrounding environment area is calculated as the ambient brightness of the pixel.

Sub-step S123: Use the ratio of the central brightness of each pixel to the ambient brightness as the feature value of the corresponding pixel.

After the central brightness and ambient brightness of each pixel are obtained, the ratio of the central brightness and ambient brightness corresponding to each pixel is calculated as the feature value of each pixel. Among them, the number of feature values in the grayscale image is equal to the number of pixels in the image. After calculating the open space feature values corresponding to each pixel point in the open space bird's eye view corresponding to the open space scene, the open space characteristic value of each pixel point may be stored in the memory in advance. By comparing the central brightness of each pixel with the ambient brightness, the interference of the illumination environment change on the obstacle recognition can be removed.

S130: Perform outlier identification processing based on the open space feature values of each pixel point of the open space bird's eye view and the actual feature value, to obtain a binary image including outliers.

It can be understood that when there is no obstacle in the actual scene, even if the overall illumination brightness changes, the ratio of the brightness between the core area of each pixel and the surrounding environment area will not change significantly. When there are obstacles in the actual scene, the characteristics of the core area and the environment area will change, and the brightness ratio of each pixel in the obstacle and the surrounding environment will change significantly. By comparing the eigenvalues of each pixel in the aerial view of the open space with the actual eigenvalues of each pixel in the actual aerial view, the abnormal points in the actual aerial view, that is, the pixels when there are obstacles on the site, can be determined. As shown in Figure 4, step S130 includes the following sub-steps:

Sub-step S131: Compare the open space feature value and the actual feature value of the pixel points at the same position to obtain a corresponding ratio.

The number of pixels in the corresponding bird's-eye view in the open-space scene and the actual scene is the same, and each pixel in the open-space bird's-eye view corresponds to the position of each pixel in the actual bird's eye view one-to-one. By comparing the actual feature values of each pixel in the actual bird's-eye view with the corresponding open-space feature values in the open-space bird's-eye view, the ratio of the feature values corresponding to each pixel can be obtained.

Sub-step S132: Determine whether the ratio is less than or equal to a preset ratio threshold.

According to the preset ratio threshold according to the actual situation, it is judged whether the ratio corresponding to each pixel point is less than or equal to the preset ratio threshold, if the ratio calculated by the pixel point is less than or equal to the preset ratio threshold, then execute sub-step S133, otherwise, execute sub-step S133 Step S134.

Sub-step S133: Determine the pixel point as the abnormal point.

Sub-step S134: Determine that the pixel point is a normal point.

If the ratio calculated by the pixel point is greater than the preset ratio threshold, it is considered that the pixel point matches, that is, the pixel point is a normal point. If the feature value of the pixel is less than or equal to the preset ratio threshold, it is determined that the pixel does not match, and the pixel is marked as an abnormal point. For example, when the preset ratio threshold is 80%, the ratio of the actual feature value of a certain pixel to the feature value of the open space is calculated to be 70%. At this time, the ratio is less than the ratio threshold, and the pixel is determined to be an abnormal point.

After determining whether each pixel in the above-mentioned actual bird's-eye view is an abnormal point, that is, after marking each abnormal point in the image, a binary image containing abnormal points can be obtained. For example, the labeling result of each pixel point can be represented by 0 and 1, and the actual bird's-eye view is marked to obtain a binary image containing abnormal points, and the binary image can be stored. Among them, 1 represents abnormal points, 0 represents normal points, at this time, black represents 0, and white represents 1.

S140: Perform denoising processing on abnormal points in the binary image to obtain a first image.

It can be understood that after the binary image is obtained, the determination result corresponding to each pixel point is a normal point or an abnormal point. When there are obstacles in the actual scene, significant abnormal points can be identified, that is, the pixels whose center brightness and ambient brightness change significantly, but because of the noise in the binary image itself, there will be a large number of discrete abnormal points. These abnormal points need to be further screened to determine whether they are pixels corresponding to obstacles. Then, in this embodiment, the pixel points marked as abnormal points in the binary image are subjected to denoising processing, that is, dilation processing and erosion processing, to obtain the first image. Among them, the abnormal points in the above binary image are denoised by the following methods:

It is preferred to perform the expansion processing of the first pixel size on the binary image to obtain the second image, which can expand the connected area of the image and prevent the noise from submerging the actual obstacles; and then perform the corrosion processing on the second image with the second pixel size to obtain the third image. image, some pixels in the second image can be removed, and some interference noises can be removed; then the third image can be expanded by the third pixel size to obtain a fourth image, and the connected areas in the third image can be expanded again; The fourth image is subjected to corrosion processing with a fourth pixel size to obtain the first image, which can restore the feature points of the binary image, and avoid the feature point size being changed by the corrosion processing and the expansion processing. The relationship between the above pixel sizes is as follows: the sum of the first pixel size and the third pixel size is equal to the sum of the second pixel size and the fourth pixel size, and the first pixel size is smaller than the second pixel size and smaller than the third pixel size .

Exemplarily, when the first pixel size is n1, the second pixel size is m1, the third pixel size is n2, and the fourth pixel size is m2, m1>n1, n2>m1, m2=n1+n2−m1, That is, n1+n2=m2+m2.

In this embodiment, the size of the expansion processing and the erosion processing are the same. Through the above-mentioned four-step denoising processing, not only the noise interference can be removed, but also the size characteristics of the binary image can be preserved.

Among them, as shown in Figure 5, the image erosion processing includes the following sub-steps:

Sub-step S141: The center point of the preset structural element is sequentially overlapped with each abnormal point in the binary image.

In this embodiment, a template matrix for the etching operation is first defined, in other words, the structural element used for the etching process is preset as the first structural element, that is, the structure used for the control operation, the center point of the first structural element is They are placed in each pixel in the above binary image marked as outliers in sequence.

Sub-step S142: Determine whether all the pixels in the structural element are abnormal points.

If all the pixels are abnormal points, the pixels corresponding to the center point are retained. Otherwise, step S143 is executed.

Sub-step S143: If all the pixels in the structural element are not all abnormal points, modify the pixels corresponding to the center point to normal points.

It is judged that all the pixels included in the first structural element are abnormal points. If the image pixels covered by all the elements in the first structural element are abnormal points, the pixels corresponding to the center point of the first structural element are reserved. Otherwise, delete the pixel point corresponding to the center point of the first structural element, that is, modify the pixel point corresponding to the center point from an abnormal point to a normal point. By sequentially overlapping the center point of the first structural element with each pixel point of the binary image, it is determined whether there is a pixel point that needs to be deleted, the main area in the image can be reduced, and the small connectivity caused by noise can be removed. area.

Exemplarily, the determination result of each pixel is represented by 0 and 1, where 0 represents a normal point and 1 represents an abnormal point. It can be defined that the size of the first structural element is 3*3 pixels and the shape is a rectangular structure, and the center point of the first structural element of 3*3 pixels is sequentially placed in the pixel point where each element is 1 in the binary image. . If the result of the pixels covered by all elements in the first structural element is 1 at this time, the pixel corresponding to the center point of the first structural element is reserved, that is, the determination result of the pixel is kept as 1; otherwise, the center point is reserved. The judgment result of the corresponding pixel is changed from 1 to 0.

The image dilation process includes the following steps:

First, a template matrix for dilation operation is defined, in other words, the structural element used for dilation processing is preset as the second structural element, that is, the structure used for control operation. For example, the size of the second structural element can be defined as 3*3 pixels, and the shape is a rectangular structure. The center point of the second structural element is coincident with each pixel point marked as an abnormal point in the binary image in turn, and it is judged whether all the pixel points covered by the second structural element are abnormal points. The normal point is modified into an abnormal point to complete the expansion action. Through the above-described expansion processing, the area of each region can be expanded, thereby filling the voids caused by noise.

S150: Perform contour detection on the first image to determine the contour of the obstacle in the actual scene.

After the first image including a plurality of connected regions is obtained, the shape features of each connected region can be identified through the contour detection process. Among them, the image contour refers to the boundary of the image, that is, the external features of the target image. As shown in Figure 6, step S150 includes the following sub-steps:

Sub-step S151: Extract at least one contour in the first image based on a preset algorithm.

In this embodiment, the image contour is obtained from the first image after denoising processing by a preset algorithm, wherein the preset algorithm may be the Satoshi Suzuki algorithm, and each image may have multiple contours, and each contour may have multiple contours. composed of multiple points.

Sub-step S152: Calculate the center point of each of the contours and the side length and area of the smallest oblique rectangle.

It can be understood that the center point and the smallest oblique rectangle corresponding to each of the above contours are determined, and the side length and area of the smallest oblique rectangle corresponding to each contour can be calculated through Python, OpenCV, etc. As shown in Figure 7, the gray part in the figure is a suspected obstacle. First, determine the minimum outline size of the obstacle. In other words, first determine the minimum resolution of the obstacle, as shown in the box in Figure 7, in which 4 obstacles have been identified. The center point of each outline is the white point mark in the box in the figure. Among them, the coordinates of the center point and the pixel size of each contour can be converted to the actual size in proportion.

Sub-step S153: Determine whether the center point of the current contour is at the boundary of the first image.

Determine whether the center point of each contour is at the border of the first image, if the center point of the current contour is not at the border of the first image, then execute sub-step S154, otherwise, execute sub-step S156, keep the current contour as Suspicious contours are used to merge with boundary images acquired by adjacent monocular cameras.

S154: Detect whether the side length is less than a preset side length threshold, and whether the area is less than a preset area threshold.

If the side length is less than the side length threshold and the area is less than the area threshold, execute sub-step S155, otherwise execute sub-step S156.

Sub-step S155: determine that the current contour is an interference contour and discard it.

Sub-step S156: Retain the current contour.

If the side length of the current contour is less than the preset side length threshold and the area is less than the preset area threshold, it is determined that the current contour is an interference contour, and the contour is discarded. Otherwise, it is determined that the contour can reflect the characteristics of the obstacle more accurately, that is, the contour of the obstacle.

Exemplarily, the preset side length threshold size is 5 pixels, and the preset area threshold size is 30 pixels. If the size of each side corresponding to the current contour is less than 5 pixels and the area is less than 30 pixels, it is considered that the current contour is Interfere with contours.

Sub-step S157: take the remaining contour as the contour of the obstacle.

The above-mentioned reserved contours are outputted, and the outputted contour can more accurately reflect the characteristics of the obstacle, that is, the above-mentioned reserved contour can be used as the contour of the obstacle.

In this embodiment, the influence of the installation angle of the monocular camera can be excluded through the perspective transformation process, and the calculation is performed on a unified top plan view, so that the size and shape features of the identified obstacles can be more accurately determined. It can remove the influence of different lighting on the recognition of image features, and can be applied to most obstacle feature recognition, avoiding the use of a large number of pre-algorithm training and testing, thereby reducing production costs and labor costs.

Based on the obstacle contour detection method of the foregoing embodiment, FIG. 8 shows a schematic structural diagram of an obstacle contour judgment apparatus 10 provided by an embodiment of the present application.

The obstacle contour judging device 10 includes:

The bird's-eye view obtaining module 11 obtains the actual image of the actual scene through the monocular camera, and performs perspective transformation on the actual image to obtain the actual bird's-eye view;

The eigenvalue calculation module 12 calculates the actual eigenvalues of each pixel in the actual bird's-eye view;

The outlier identification module 13 performs outlier identification processing based on the open space characteristic value of each pixel point of the open space bird's eye view and the actual characteristic value, and obtains a binary image containing outliers;

The denoising processing module 14 performs denoising processing on the abnormal points in the binary image to obtain a first image;

The contour determination module 15 performs contour detection on the first image to determine the contour of the obstacle in the actual scene.

The present embodiment provides an obstacle contour determination device 10, which is used to implement the above implementation through the cooperative use of a bird's-eye view acquisition module 11, a feature value calculation module 12, an abnormal point identification module 13, a denoising processing module 14, and a contour determination module 15. For the obstacle contour detection method described in the example, the implementation and beneficial effects involved in the above-mentioned embodiment are also applicable in this embodiment, and are not repeated here.

In addition, the present application also provides a terminal device, including a memory and a processor, wherein the memory stores a computer program, and the computer program executes the obstacle contour detection method described in the foregoing embodiments when running on the processor.

This embodiment also provides a computer storage medium for storing the computer program used in the above-mentioned terminal device.

In the several embodiments provided in this application, it should be understood that the disclosed apparatus and method may also be implemented in other manners. The apparatus embodiments described above are only schematic, for example, the flowcharts and structural diagrams in the accompanying drawings show the possible implementation architectures and functions of the apparatuses, methods and computer program products according to various embodiments of the present invention and operation. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code that contains one or more functions for implementing the specified logical function(s) executable instructions. It should also be noted that, in alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It is also noted that each block of the block diagrams and/or flowchart illustrations, and combinations of blocks in the block diagrams and/or flowchart diagrams, can be implemented using dedicated hardware-based systems that perform the specified functions or actions. be implemented, or may be implemented in a combination of special purpose hardware and computer instructions.

In addition, each functional module or unit in each embodiment of the present invention may be integrated together to form an independent part, or each module may exist alone, or two or more modules may be integrated to form an independent part.

If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such understanding, the technical solution of the present invention can be embodied in the form of a software product in essence, or the part that contributes to the prior art or the part of the technical solution. The computer software product is stored in a storage medium, including Several instructions are used to cause a computer device (which may be a smart phone, a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes .

The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited to this. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention. should be included within the protection scope of the present invention.

Claims (10)

1. an obstacle contour detection method, is characterized in that, comprises: Obtain an actual image of the actual scene through a monocular camera, and perform perspective transformation on the actual image to obtain an actual bird's-eye view; Calculate the actual feature value of each pixel in the actual bird's-eye view; Perform outlier identification processing based on the open space feature value of each pixel point of the open space bird's-eye view and the actual characteristic value, and obtain a binary image containing outliers; Denoising the abnormal points in the binary image to obtain a first image; Perform contour detection on the first image to determine the contour of the obstacle in the actual scene. 2. The obstacle contour detection method according to claim 1, wherein the performing contour detection on the first image to determine the contour of the obstacle in the actual scene comprises: extracting at least one contour in the first image based on a preset algorithm; Calculate the center point of each of the contours and the side length and area of the smallest oblique rectangle; If the center point of the current contour is not at the boundary of the first image, detecting whether the side length is less than a preset side length threshold, and whether the area is less than a preset area threshold; If the side length is less than the side length threshold and the area is less than the area threshold, then the current contour is determined to be an interference contour and discarded, otherwise the current contour is retained; If the center point of the current contour is at the boundary of the first image, retaining the current contour; Take the remaining contour as the contour of the obstacle. 3. The obstacle contour detection method according to claim 1 or 2, characterized in that, before obtaining the actual image of the actual scene by the monocular camera, the method comprises: Obtain an open space image of the open space scene through a monocular camera, and measure the actual size of the open space scene to determine the aspect ratio of the open space scene; Determine the coordinates of the open space based on the aspect ratio and the resolution of the open space aerial view of the open space scene to be generated; Setting a corner point at the boundary of the open space scene, and determining the corner point coordinates corresponding to the corner point in the open space image; Calculate the parameters of the perspective transformation matrix based on the corner coordinates and the open space coordinates to determine the perspective transformation matrix; Perspective transformation is performed on the open space image through the perspective transformation matrix to obtain an open space bird's-eye view of the open space scene. 4. obstacle contour detection method according to claim 1 is characterized in that, the eigenvalue of each pixel point in the corresponding bird's-eye view is calculated by the following manner: Perform grayscale processing on the corresponding bird's-eye view to obtain a grayscale image; Calculate the corresponding center brightness and ambient brightness based on the respective pixel points of the grayscale image as the center; According to the ratio of the central brightness of each pixel to the ambient brightness, the feature value of the corresponding pixel is taken. 5. The obstacle contour detection method according to claim 4, characterized in that, carrying out outlier identification processing based on the open space feature value of each pixel point of the open space bird's eye view and the actual feature value, comprising: Comparing the open space feature value and the actual feature value of the pixel point at the same position to obtain a corresponding ratio; If the ratio is less than or equal to a preset ratio threshold, the pixel point is determined to be the abnormal point; otherwise, the pixel point is determined to be a normal point. 6 . The obstacle contour detection method according to claim 1 , wherein, performing denoising processing on the abnormal points in the binary image to obtain the first image, comprising: 6 . Dilation processing of the first pixel size is performed on the binary image to obtain a second image; Erosion processing of the second pixel size is performed on the second image to obtain a third image; Dilation processing of a third pixel size is performed on the third image to obtain a fourth image; Erosion processing of a fourth pixel size is performed on the fourth image to obtain a first image; The relationship between the first pixel size, the second pixel size, the third pixel size and the fourth pixel size is as follows: the sum of the first pixel size and the third pixel size is equal to the The sum of the second pixel size and the fourth pixel size, the first pixel size is smaller than the second pixel size is smaller than the third pixel size. 7. The obstacle contour detection method according to claim 6, wherein the corrosion processing comprises: The center point of the preset structural element is sequentially overlapped with each abnormal point in the binary image, and it is judged whether all the pixel points in the structural element are abnormal points; If all the pixel points are the abnormal points, keep the pixel points corresponding to the center point; Otherwise, modify the pixel point corresponding to the center point to a normal point. 8. A device for judging the contour of an obstacle, comprising: The bird's-eye view acquisition module obtains the actual image of the actual scene through the monocular camera, and performs perspective transformation on the actual image to obtain the actual bird's-eye view; an eigenvalue calculation module, which calculates the actual eigenvalues of each pixel in the actual bird's-eye view; The outlier identification module performs outlier identification processing based on the open space characteristic value of each pixel point of the open space bird's eye view and the actual characteristic value, and obtains a binary image containing outliers; a denoising processing module, which performs denoising processing on the abnormal points in the binary image to obtain a first image; The contour determination module performs contour detection on the first image to determine the contour of the obstacle in the actual scene. 9. A terminal device, characterized by comprising a memory and a processor, wherein the memory stores a computer program, and the computer program executes the obstacle according to any one of claims 1 to 7 when the computer program runs on the processor Object contour detection method. 10 . A readable storage medium, characterized in that it stores a computer program, and the computer program executes the obstacle contour detection method according to any one of claims 1 to 7 when the computer program runs on a processor.

Images