KR101212127B1 - Method and apparatus for detecting obstacle - Google Patents

Abstract

Obstacle object detection device based on the parallax map (depth map) and the camera information, calculates the position of the obstacle object, the bird's eye view mapping unit for generating a bird's eye view of the obstacle object by displaying the position of the obstacle object on the plane Grouping unit for grouping row by row, and generating disparity map histogram for each group for bird's eye view, detecting peak and valley on histogram, and area of obstacle object for each group using peak and valley It includes a clustering unit for detecting.

Description

Obstacle object detection device and method {METHOD AND APPARATUS FOR DETECTING OBSTACLE}

The present invention relates to an obstacle object detection apparatus and method.

Obstacles are detected using a parallax map, and there are a method using a V-disparity map and a method using column detection.

1 is a diagram illustrating a result of detecting an obstacle object using a conventional stereo vision on a parallax map.

Referring to FIG. 1, a plurality of obstacle objects may be detected as one obstacle object, depending on the resolution of the parallax map or the proximity of the adjacent obstacle object. There are two obstacle object areas detected in FIG. 1. However, two obstacle objects 112 and 114 exist in the first obstacle object area 110, and three obstacle objects 122, 124 and 126 exist in the second obstacle object area 120. Since the two obstacle objects 112 and 114 and the three obstacle objects 122, 124 and 126 are obstacle objects located at close distances, respectively, they are detected as one obstacle object. Therefore, it is necessary to accurately detect the obstacles.

Disclosure of Invention The technical problem to be solved by the disclosed technology is to provide an obstacle object detection apparatus and method for precisely detecting an obstacle object by generating a bird's eye view on the obstacle, performing grouping and clustering on the bird's eye view, and performing an inverse bird's eye mapping. There is.

In order to achieve the above technical problem, a first aspect of the disclosed technology calculates a position of an obstacle object based on a depth map and camera information, and displays the position of the obstacle object on a plane to display the object. A bird's eye view mapping unit for generating a bird's eye view, a grouping unit for grouping the bird's eye view by row, and a parallax map histogram for each group with respect to the bird's eye view, and detect peaks and valleys on the histogram The present invention provides a obstacle object detecting apparatus including a clustering unit for detecting an area of an obstacle object for each group by using the peak and valley.

In order to achieve the above technical problem, a second aspect of the disclosed technology calculates a position of an obstacle object based on a difference map and camera information, and displays the position of the obstacle object on a plane to display it on the obstacle object. A bird's eye view mapping unit for generating a bird's eye view, a grouping unit for grouping the bird's eye view at a predetermined interval according to the front distance of the disparity map, and performing clustering on the bird's eye view for each group of obstacle object areas for each group It provides a obstacle object detection apparatus comprising a clustering unit for detecting a; and an inverse bird's-eye mapping unit for performing an inverse bird's-eye mapping on the area of the obstacle object detected by the clustering unit to represent the area of the obstacle object as a parallax map. There is.

The third aspect of the disclosed technology to achieve the above technical problem is to calculate the position of the obstacle object based on the difference map and camera information, and to display the position of the obstacle object on the plane to the obstacle object Generating a bird's-eye view, grouping row by row for the bird's-eye view, and generating a disparity map histogram for each group for the bird's-eye view, detecting peaks and valleys on the histogram, and The present invention provides a method for detecting an obstacle object, the method including detecting an area of an obstacle object for each group using a peak and a valley.

In order to achieve the above technical problem, a fourth aspect of the disclosed technology calculates a position of an obstacle object based on a difference map and camera information, and displays the position of the obstacle object on a plane to display it on the obstacle object. Generating a bird's eye view for the bird's eye view, grouping the bird's eye bird's eye view at a predetermined interval according to the distance of the disparity map, and performing clustering on the bird's eye view to detect the obstacle object region for each group. And performing a reverse bird's-eye mapping on the area of the obstacle object detected by the clustering unit to represent the area of the obstacle object as a parallax map.

The disclosed technique may have the following effects. It is to be understood, however, that the scope of the disclosed technology is not to be construed as limited thereby, as it is not meant to imply that a particular embodiment should include all of the following effects or only the following effects.

The obstacle detecting apparatus according to an embodiment generates a bird's eye view of the obstacle object even in a situation where it is difficult to accurately detect the obstacle object due to the resolution of the parallax map or the proximity of the adjacent obstacle object, and grouping and clustering on the bird's eye view By performing the method and performing the reverse bird's eye mapping, it is possible to accurately detect the obstacle. In addition, the obstacle detecting apparatus according to an embodiment may be applied not only to moving objects such as intelligent cars or robots, but also to various everyday spaces such as factories and homes.

1 is a diagram illustrating a result of detecting an obstacle object using a conventional stereo vision on a parallax map.
2 is a block diagram showing the configuration of an obstacle detecting apparatus according to an embodiment of the present invention.
FIG. 3 is a diagram for describing an example in which the obstacle detecting apparatus of FIG. 2 detects an obstacle object.

The description of the disclosed technique is merely an example for structural or functional explanation and the scope of the disclosed technology should not be construed as being limited by the embodiments described in the text. That is, the embodiments may be variously modified and may have various forms, and thus the scope of the disclosed technology should be understood to include equivalents capable of realizing the technical idea.

Meanwhile, the meaning of the terms described in the present application should be understood as follows.

Singular expressions should be understood to include plural expressions unless the context clearly indicates otherwise, and terms such as "include" or "have" refer to features, numbers, steps, operations, components, parts, or parts thereof described. It is to be understood that the combination is intended to be present, but not to exclude in advance the possibility of the presence or addition of one or more other features or numbers, steps, operations, components, parts or combinations thereof.

Each step may occur differently from the stated order unless the context clearly dictates the specific order. That is, each step may occur in the same order as described, may be performed substantially concurrently, or may be performed in reverse order.

All terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the disclosed technology belongs, unless otherwise defined. Terms defined in commonly used dictionaries should be interpreted to be consistent with meaning in the context of the relevant art and can not be construed as having ideal or overly formal meaning unless expressly defined in the present application.

2 is a block diagram showing the configuration of an obstacle detecting device according to an embodiment of the present invention. Referring to FIG. 2, the obstacle detecting apparatus 200 includes a bird's eye view mapping unit 210, a grouping unit 220, a clustering unit 230, an inverse bird's-eye view mapping unit 240, a quantization error removing unit 250, and filtering. The unit 260 is included.

The bird's eye view mapping unit 210 generates a bird's eye view based on a parallax map and camera information. The parallax map may be generated as a result of stereo matching on the left image and the right image. The camera information may include information about a separation distance between the cameras generating the left image and the right image, and angles at which the cameras view the ground. The bird's eye view mapping unit 210 may generate a bird's eye view by calculating a position of the obstacle object based on the parallax map and the camera information and displaying the position of the obstacle object on a plane. Here, the bird's eye view does not display the height of the obstacle object in nature, and may display a horizontal position and a front position (ie, distance).

The grouping unit 220 performs grouping on a row-by-row basis for the bird's eye view. The grouping unit 220 may perform grouping on rows having a predetermined interval with respect to the front distance. Here, the grouping interval may be changed according to the separation performance of the obstacle. For example, the narrower the grouping row spacing, the better the separation performance of the obstacle and the slower it may be. As the grouping row spacing becomes wider, the speed may be faster and the separation performance of the obstacle may be worse. Through this, after the bird's eye view mapping, it is possible to remove the noise due to the error of the parallax map, it is possible to improve the speed.

The clustering unit 230 performs clustering for each group with respect to the bird's eye view. The clustering unit 230 generates a disparity map histogram for each group, and detects a peak and a valley on each histogram. The clustering unit 230 detects the area of the obstacle object for each group on the bird's eye view based on the detected peak and valley.

The inverse bird's-eye view mapping unit 240 maps the region of the obstacle object detected by the clustering unit 230 to the inverse bird's-eye view and represents the disparity map. However, due to quantization, the obstacle object region may appear as a straight line or a dotted line when mapping the reverse bird's-eye view.

The quantization error remover 250 scans rows by row within a limited horizontal position based on the inverse bird-eye mapped value to detect pixel positions having similar values. That is, if the obstacle object is a plane and has similar parallax values in the plane, most parallax values of the obstacle object have values similar to the reverse bird's eye mapped values. Through this, the quantization error remover 250 removes the quantization error generated at the inverse bird's eye mapping.

The filtering unit 260 detects only a desired obstacle object by performing noise removing or size limit filtering.

FIG. 3 is a diagram for describing an example in which the obstacle detecting apparatus of FIG. 2 detects an obstacle object.

Referring to (a) of FIG. 3, there are two obstacle regions composed of the first obstacle object region 310 and the second obstacle region 2 320 on the parallax map, and the first obstacle object region 310 is located in the parallax map. There are two obstacles 312 and 314, and there are three obstacles 322, 324 and 326 in the second obstacle object area 320.

The bird's eye view mapping unit 210 performs bird's-eye view mapping based on the camera information and the parallax map, and identifies five obstacles for each horizontal position X and the front distance Y, as shown in FIG. .

As shown in (c) of FIG. 3, the grouping unit 220 performs grouping on rows of a predetermined distance in front of the distance Y. Here, the grouping row spacing may be changed according to the separation performance of the obstacle. For example, the narrower the grouping row spacing, the better the separation performance of the obstacle and the slower it may be. As the grouping row spacing becomes wider, the speed may be faster and the separation performance of the obstacle may be worse.

The clustering unit 230 generates a disparity map histogram for each group, and detects peaks and valleys on each histogram. The clustering unit 230 detects regions of the obstacle object for each group on the bird's eye view using peaks and valleys as shown in FIG.

The inverse bird's-eye view mapping unit 240 maps the region of the obstacle object detected by the clustering unit 230 to inverse bird's-eye view and expresses the image on the parallax map as shown in FIG.

As illustrated in (f) of FIG. 3, the quantization error remover 250 detects pixel positions having similar values by scanning rows by row within a limited horizontal position based on the inverse bird-eye mapped value. That is, if the obstacle object is a plane and has similar parallax values in the plane, most parallax values of the obstacle object have values similar to the reverse bird's eye mapped values.

The filtering unit 260 detects only a desired obstacle object as shown in FIG. 3G by performing noise removing or size limit filtering.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the present invention as defined by the following claims It can be understood that

Claims (12)

On the basis of a parallax map and information about a camera generating a left image and a right image for creating the parallax map, a position of the obstacle object is calculated, and the position of the obstacle object is displayed on a plane to prevent the obstacle. A bird's eye view mapping unit for generating a bird's eye view of the object;
A grouping unit to group the bird's eye by row; And
Clustering for generating a disparity map histogram for each of the groups for the bird's eye view, detecting peaks and valleys on the histogram, and detecting an area of an obstacle object for each group using the peaks and valleys Obstacle object detection device comprising a. The method of claim 1, wherein the grouping unit
Obstacle object detecting apparatus for performing grouping at regular intervals in accordance with the front distance of the parallax map. The method of claim 2,
And an inverse bird's-eye mapping unit configured to perform inverse bird's-eye mapping on the area of the obstacle object detected by the clustering unit to represent the area of the obstacle object as a parallax map. The method of claim 3, wherein
And a quantization error canceling unit configured to scan pixel by row in a horizontal position of the obstacle object region represented by the inverse bird's eye view and detect a pixel position having a value within a preset range. On the basis of a parallax map and information about a camera generating a left image and a right image for creating the parallax map, a position of the obstacle object is calculated, and the position of the obstacle object is displayed on a plane to prevent the obstacle. A bird's eye view mapping unit for generating a bird's eye view of the object;
A grouping unit that performs grouping at a predetermined interval with respect to the bird's eye view according to the distance of the disparity map;
Clustering for generating a disparity map histogram for each of the groups for the bird's eye view, detecting peaks and valleys on the histogram, and detecting an area of an obstacle object for each group using the peaks and valleys part; And
And an inverse bird's-eye mapping unit configured to perform inverse bird's-eye mapping on the area of the obstacle object detected by the clustering unit to represent the area of the obstacle object as a parallax map. delete The method of claim 5, wherein
And a quantization error canceling unit configured to detect pixel positions having a value within a preset range by scanning row by row within a horizontal position of the obstacle object region mapped to the inverse bird's eye view. On the basis of a parallax map and information about a camera generating a left image and a right image for creating the parallax map, a position of the obstacle object is calculated, and the position of the obstacle object is displayed on a plane to prevent the obstacle. Generating a bird's eye view of the object;
Grouping row by row for the bird's eye view; And
Generating a disparity map histogram for each group with respect to the bird's eye view, detecting a peak and a valley on the histogram, and detecting an area of an obstacle object for each group using the peak and the valley Obstacle object detection method comprising a. The method of claim 8, wherein the grouping step
The obstacle object detection method for performing grouping at regular intervals in accordance with the front distance of the parallax map. The method of claim 9,
And performing an inverse bird's-eye mapping on the detected obstacle object area to represent the area of the obstacle object as a parallax map. 11. The method of claim 10,
And detecting a pixel position having a value within a preset range by scanning row by row within a horizontal position of the obstacle object region represented by the reverse bird's-eye view. On the basis of a parallax map and information about a camera generating a left image and a right image for creating the parallax map, a position of the obstacle object is calculated, and the position of the obstacle object is displayed on a plane to prevent the obstacle. Generating a bird's eye view of the object;
Performing grouping at a predetermined interval with respect to the bird's eye view according to the front distance of the parallax map;
Generating a parallax map histogram for each of the groups with respect to the bird's eye view, detecting peaks and valleys on the histogram, and performing clustering for detecting an area of an obstacle object for each group using the peaks and valleys; And
And performing an inverse bird's-eye mapping on the area of the obstacle object detected by the clustering to represent the area of the obstacle object as a parallax map.

Images