JP2002329202A - Object detecting device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an object detecting in which detected object information is subjected to decision processing with high-reliability. SOLUTION: This device is equipped with an infrared camera 1, a storage means which stores 1st object information corresponding to shape characteristics of a human body 8 having heating generating characteristics and 2nd and 3rd pieces of object information corresponding to shape characteristics of a background, etc., and an image processing means 3 which determines decision object information 14 in image information picked up by the infrared camera and performs a 1st decision processing for deciding which of the preciously stored 1st and 2nd object information the decision object information 14 approximates to through comparison by a characteristic space method and then a 2nd decision processing for deciding which of the 1st and 3rd pieces of object information the decision object information judged by the 1st decision making process to approximate to the 1st object information by comparison by the characteristic space method.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an object detecting device mounted on a vehicle or the like, and more particularly, to an object detecting device provided with an infrared type photographing means.
The present invention relates to an object detection device that accurately detects a heat-generating object such as a human or an animal.

[0002]

2. Description of the Related Art As an object detecting device, there is a human body detecting device which detects a person (human body) in front of a vehicle and alerts a driver, and there is one disclosed in Japanese Patent Application Laid-Open No. 9-326096.

This human body detection device includes an infrared camera. In an infrared image input from the infrared camera, a pixel group whose luminance value is within a specific range is left and a connected pixel group is obtained, and the shape of the pixel group is determined. Connected pixel group)
Is a person whose shape is circular and whose shape is a specific size.

[0004]

In such a human body detecting device, there is a possibility that a heat generating object other than a person close to the temperature or the size of the person may be determined as a person, and a highly reliable human body determination can be performed. There is a problem that the commercial value of the human body detecting device is lowered.

Accordingly, the present invention focuses on the above-mentioned problems, and provides an object detecting device using infrared photographing means.
It is an object of the present invention to provide an object detection device capable of performing highly reliable determination processing on image information from a photographing unit.

[0006]

In order to solve the above-mentioned problems, the object detecting apparatus of the present invention has an infrared photographing means for photographing a scene in front of a vehicle and has an exothermic characteristic. Storage means for storing first object information corresponding to the shape characteristics of the first object and second and third object information corresponding to the shape characteristics of the second and third objects other than the first object Determining the information to be determined in the image information picked up by the photographing means, and determining which of the first object information and the second object information stored in advance is closer to the determined information. After performing a first determination process for comparison based on the eigenspace method, the first determination process is performed on the determined information approximated to the first object information in the first determination process.
Control means for detecting the first object in the image information by performing a second determination process of comparing which of the object information and the third object information is closer to each other based on the eigenspace method. It is characterized by comprising.

[0007] Also, as described in claim 2, claim 1
In the object detection device according to the above, the control unit extracts a high-temperature portion of the heat-generating object from the image information, classifies the high-temperature portion into a region having a predetermined size, Information is extracted as the information to be determined.

[0008] Also, as described in claim 3, claim 1
Alternatively, in the object detection device according to claim 2, the storage unit stores a human body determination result of the previous determination target information in the image information captured by the imaging unit, and the control unit newly performs image capturing. A third determination process for determining whether or not the object is the first object is performed based on at least the previous human body determination result in the current determination target information obtained as described above. .

Further, as described in claim 4, claim 1 is
4. The object detection device according to claim 3, further comprising a display unit configured to display the image information, wherein the control unit controls the display unit to highlight the shape of the first object. It is characterized by becoming.

[0010] Further, as described in claim 5, claim 1 is
5. The object detection device according to claim 1, wherein the storage unit stores the first and second shapes according to a shape characteristic of a person or an animal.
In which the object information is stored.

[0011]

Embodiments of the present invention will be described below with reference to the accompanying drawings.

Hereinafter, a human body detecting device mounted on a vehicle will be described as an example of the object detecting device. In FIG.
The human body detection device 1 is an infrared camera (infrared imaging means)
2, image processing means (control means) 3, image display means 4
And an alarm means 5.

The infrared camera 2 is disposed on a front grill, a front bumper, and the like in a vehicle (for example, an automobile) 6. As shown in FIG. 2, the infrared camera 2 has an imaging area 7 composed of, for example, 240 pixels × 320 pixels.

The image processing means 3 detects a human body (heating object) 8 located at a predetermined distance from the vehicle 6 and will be described later based on image data (image information) based on the imaging area 7 of the infrared camera 2. This is control means for detecting a heat-generating object as the human body 8 by processing, and is constituted by a microcomputer.

The image display means 4 is, for example, a so-called head-up display which is constituted by a liquid crystal display device with a backlight and projects an image of a human body 8 as display light from the liquid crystal display device onto a windshield 9. The driver 10 in the vehicle 6 can visually recognize the image without deviating his / her eyes from the scenery ahead.

The alarm means 5 is constituted by a lamp, an LED, a buzzer and the like. When the image processing means 3 determines that the human body 8 exists in the imaging area 7 of the infrared camera 2, it outputs a predetermined drive signal to the alarm means 5 to operate.

The above components constitute the human body detecting device 1.

Next, a processing procedure in the image processing means 3 of the human body detecting device 1 will be described with reference to FIGS.

The image processing means 3 inputs image information based on the imaging area 7 from the infrared camera 2 (step S).
1).

The image processing means 3 executes a binarization process on a predetermined range of image information detected by the infrared camera 2, and is a first heat-generating region having a temperature higher than the ambient temperature.
(High-temperature portion) 11 and second image data 12 serving as a non-heat-generating region (step S2).

Next, the image processing means 3 groups the first image data 11 to be connected on a pixel-by-pixel basis to determine the size of the first image data 11, and A surrounding area 13 (for example, a circumscribed rectangle) that contacts the contour of the image data 11 is generated (Step S)
3). FIG. 4 shows an extraction result of the image processing means 3.

The image processing means 3 performs a determination assuming that a scene that becomes a noise is included in the determination of the size of the first image data. Is a determination to remove the first image data 11 that is a small area of 4 pixels or less and the first image data 11 that is a rectangular area whose long side of the surrounding area 13 is 32 pixels or more. That is, in the first image data of the surrounding area 13, the first image data 11 having a size clearly outside the determination target can be removed from the object having a temperature higher than the ambient temperature.

In the above-described determination, the image processing means 3 determines that only the surrounding region 13 in which the length X of the long side of the surrounding region 13 is a value larger than 4 pixels and smaller than 32 pixels (4 <X <32) is used as a human body. 8 and the surrounding area 13 not included in this range is removed (step S4).
The number of pixels related to the removal of the minute area and the rectangular area differs depending on the distance from the infrared camera 2 to the human body 8, the pixel configuration of the infrared camera 2, and the like.

Next, in step S2, the image processing means 3 calculates the center of gravity of the connected component in the surrounding area 13 which has not been removed, and centers the center of gravity on this connected center, for example, as shown in FIG. By generating a cutout area of 32 pixels × 32 pixels, the determined information 14 is set,
It counts how many pieces of the information to be determined 14 are included in the image information in the imaging area, and inputs the number of pieces of the information to be determined 14 (the number of times of determination) for performing the human body selection determination described below to N. At the same time, the count value of the number of executions n is initialized (n ← 0) (step S5).

Next, the image processing means 3 determines whether or not all of the information to be determined 14 has been subjected to human body selection determination by comparing the number of information to be determined N with the number of executions n (step S6). If n does not reach the number N of information to be determined, the human body selection determination processing from step S6 to step S19 is executed.

The image processing means 3 determines whether or not the object corresponding to the determination target information 14 has been subjected to the human body selection determination at the time of the previous input of the image information (step S7). In this case, the image processing means 3 inputs the current image information into the determined information 14 obtained when the previous image information was input, and proceeds to step S
This is to determine whether or not the storage means described later stores position data that is the same as or close to the position of the determination target information 14 set in step 5. Further, the image processing means 3
In step S7, if the to-be-determined information 14 has been subjected to the human body selection determination processing last time, the value of the human body determination counter (human body determination result) C of the previous to-be-determined information 14 is changed to the human body determination of the current to-be-determined information 14. It is input to the number of times C (step S8), and in step S6, if the to-be-determined information 14 has not been subjected to the human body selection determination processing last time, the counter value of the human body determination counter C is initialized (C ← 0). (Step S
9).

Next, the image processing means 3 outputs the information to be determined 14
Are at least one eigenspace (hereinafter, referred to as first object information) representing features of the human body 8 stored in advance, and at least one eigenspace (hereinafter, second object) representing features of the background (landscape). (Hereinafter referred to as information) is determined (step S10).

The determination process (first determination process) in step S10 will be described with reference to FIG. In the following description, the eigenspace is represented two-dimensionally.

The image processing means 3 stores information to be determined 14 including information such as image data, position data, size, and the number of times of human body determination C in a ROM or EE provided in the image processing means 3.
To determine whether the first object information P1 for human body recognition or the second object information P2 for background recognition previously stored in a storage means (not shown) such as a PROM or the like is similar. It is.

That is, the image processing means 3 determines the magnitudes of the first and second vectors p1 and p2 when projecting the information to be determined 14 onto the first and second object information P1 and P2 (lengths in FIG. Is determined by a predetermined calculation, and the magnitudes of the vectors p1 and p2 are compared, so that the determined information 14 is the first object information P1 for human body recognition and the second object information P1 for background recognition. P2 is determined as to which image data the image data is closer to, and it is determined whether the image data is the human body 8 or the background. The object information P1 and P2 stored in the image processing means 3 in advance are the eigenspace of the human body 8 and the background at a position where the distance between the vehicle 6 and the human body 8 is a predetermined distance (for example, 80 m).

Further, the image processing means 4 determines in step S10
In the determination process of the above, when it is determined that the determined information 14 is close to the first object information, that is, it has a shape characteristic close to a human body, the determined information 14 is stored in the human body 8 stored in advance.
And at least one eigenspace (hereinafter, referred to as third object information) representing characteristics of an object other than the first and second object information. Is determined (step S
11). Note that, in this case, the image processing means 3 preliminarily stores the shape characteristics of the traffic light, the electric pole and the like as the third object information in the storage means.

The determination processing (second determination processing) in step S11 is the same as the processing method in step S10, and thus a detailed description is omitted.

Next, when the image processing means 3 determines in the determination processing in step S11 that the information to be determined 14 has a shape characteristic close to the first object information, that is, has a shape characteristic close to the human body 8, the image processing means 3 The count value is incremented (C ← C + 1) (step S12).

Next, the image processing means 3 stores the image data, the position data, the size, and the number of times of human body determination C of the information to be determined 14 in the storage means (step S13). At the time of inputting the image information twice, it is determined whether or not the human body selection has been determined (step S14). In this case, the image processing means 3 inputs the current image information into the determined information 14 obtained when the past three times of the image information are input, extracts the image information in the step S4, and stores the image information in the step S13. This is to determine whether or not there is any data having position data close to the position of the determination information 14.

In step S14, if the human body selection has not been determined in the past three times of inputting the image information for the same object, the object corresponding to the determined information 14 is a candidate for the human body 8 in step S14. (Step S
15). If the object corresponding to the information 14 to be determined has been selected three or more times in the past for the object corresponding to the information 14 to be determined, the image processing means 3 performs the third determination process as the third determination process.
It is determined whether or not the number of human body determinations C is 2 or more (step S16). If the number of human body determinations C is 2 or more, it is recognized that the object corresponding to the determined information 14 is the human body 8 (step S17). If the number of times of human body determination C is less than 2, the object corresponding to the information to be determined 14 is recognized as an object other than the human body 8.

The image processing means 3 increments the number of executions n (n ← n + 1) (step S19). In step S6, when comparing the number of information to be determined N with the number of executions n, the information to be determined 14 Are repeated for the to-be-determined information 14 in the image information, and the human-body selection is determined for all of the to-be-determined information 14 in the image information until it is determined that all of the information has been determined. After that, the processing result of the image information by the image processing means 3 is notified to the driver using the image display means 4 and the alarm means 5 (step S20). Step S
When performing the process of step 20, the image information unit 3 executes step S1.
In the case where the image information input in step S17 includes the determined information 14 recognized as the human body 8 in step S17 in the image information in the image pickup area 7, the image display unit 4 is controlled to control the image display means 4 as shown in FIG. As shown by, only the portion corresponding to the information to be determined, that is, only the human body 8 portion is displayed, and the human body 8 in the imaging area 7 is controlled by controlling the alarm means 5.
Is positively notified to the driver 10.

The human body detecting device 1 includes an infrared camera 1 for photographing a scene in front of a vehicle 6, a first object information corresponding to a shape characteristic of a human body 8 having a heat generation characteristic, and a second object information corresponding to a shape characteristic of a background. Storage means for storing the third object information corresponding to the shape information of the object other than the above object information, and the determined information 14 in the image information captured by the infrared camera 1, and the determined information 14 On the other hand, after performing a first determination process of comparing which of the first object information and the second object information stored in advance is closer to each other based on the eigenspace method, in the first determination process, A second determination process of comparing which of the first object information and the third object information is closer to the determined information approximated to the first object information based on the eigenspace method; By doing so, the image information And the image processing means 3 for detecting the human body 8 in the inside, the erroneous determination can be prevented, and the object detection that can perform highly reliable determination processing on the detected object information An apparatus is provided.

The storage means stores the result of the previous human body determination of the to-be-determined information 14 in the image information picked up by the infrared camera 1, and the image processing means 3 stores the newly obtained image picked up By performing a third determination process of determining whether or not the subject is the human body 8 based on at least the previous human body determination result in the determined information 14, the image processing unit 3 The first and second determination processes are performed a plurality of times until the human body 8 is recognized without determining the human body 8 with only one piece of image information, so that a more reliable determination result can be obtained.

The image processing means 3 comprises an image display means 4
In the display method of the human body 8 described above, since the determined information 14 to be a cut-out area is not displayed on the image display means 4 but image data of only the human body 8 in the determined information 14 is displayed, the binarization is performed. Human body 8 obtained by processing
Can be displayed as it is, and the image display means 4 can display the human body 8 with emphasis.

The provision of the warning means 5 such as an LED, a lamp and a buzzer enables the driver 10 to easily determine the human body 8 existing in front of the vehicle 6.

In the embodiment of the present invention, the human body detecting device 1 for detecting the human body 8 has been described. However, the data of the eigenspace representing the characteristics of animals such as dogs and cats and the characteristics other than the animals are described. Is stored in advance, it can be used as an object detection device that can also detect a heating object other than the human body 8.

In the embodiment of the present invention, one eigenspace for human body recognition and one eigenspace for background recognition are compared in the description of FIG. Since various states such as a horizontal state and a sitting state are assumed, and there are various states in the background, a plurality of eigenspaces for human body recognition and a plurality of eigenspaces for background recognition are stored. Needless to say, you need to do that.

Further, in the embodiment of the present invention, the above-described processing determination is performed on the image data in the entire area of the imaging area 7. However, the present invention can best capture a heating object that is separated by a predetermined distance. The processing speed in the image processing unit 3 can be improved by generating a predetermined partitioned area in the possible area and performing the determination process on the partitioned area.

In the embodiments of the present invention, a head-up display is taken as an example of the display means. However, for example, a liquid crystal display device may be used, and the display means in the present invention is limited to the embodiments of the present invention. It is not done.

In the embodiment of the present invention, only the portion of the human body 8 detected by the image display means 4 is displayed. However, the present invention is not limited to this. 4, the inside of the imaging area is displayed as an image including the background, and the human body 8 can be emphasized by coloring the human body 8 in red or the like. In addition, by displaying the object (candidate of the human body) corresponding to the determination target information 14 processed in step S15 in yellow or the like, a further warning display in the human body detection is enabled.

[0046]

The present invention relates to an object detecting device mounted on a vehicle or the like, and more particularly to an object detecting device provided with an infrared type photographing means for detecting a heating object with high accuracy. Object detection device capable of performing a high-accuracy determination process.

[Brief description of the drawings]

FIG. 1 is a block diagram of a human body detection device according to an embodiment of the present invention.

FIG. 2 is a diagram showing an image photographed by the imaging means of the embodiment.

FIG. 3 is an exemplary view showing a processing procedure by the image processing means of the embodiment.

FIG. 4 is a view showing an image processed by a first determination process of the embodiment.

FIG. 5 is a view showing third image data according to the embodiment;

FIG. 6 is an exemplary view for explaining a second processing determination in the embodiment.

FIG. 7 is a diagram showing a display state of an image display unit after a second processing determination in the embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Human body detection device (object detection device) 2 Infrared camera (infrared imaging means) 3 Image processing means (control means) 4 Image display means (display means) 6 Vehicle 8 Human body (first object) 11 First image data (High temperature part) 14 Information to be determined P1 Eigenspace for human body recognition (first object information) P2 Eigenspace for background recognition (second object information)

Continued on the front page F term (reference) 5B057 AA16 BA02 CA12 CA16 DA07 DA15 DB02 DC03 DC09 DC36 5H180 AA01 CC02 CC04 FF33 LL01 LL07 LL08 5L096 BA02 BA04 CA02 DA03 FA02 FA59 FA64 JA11

Claims (5)

[Claims] An infrared photographing means for photographing a scene in front of a vehicle; first object information corresponding to a shape characteristic of a first object having a heat generation characteristic; Storage means for storing second and third object information corresponding to the shape characteristics of the third object, and information to be determined in image information taken from the photographing means;
A first method for comparing which of the first object information and the second object information stored in advance is closer to each other based on the eigenspace method;
After performing the determination process of the above, in the first determination process, for the determined information approximated to the first object information,
The first object in the image information is detected by performing a second determination process of comparing which of the first object information and the third object information approximates based on the eigenspace method. An object detection device, comprising: a control unit. 2. The control means extracts a high-temperature portion of the heat-generating object from the image information, classifies the high-temperature portion having a predetermined size into regions, and extracts information in the region. The object detection device according to claim 1, wherein the object detection device is extracted as information to be determined. 3. The storage means stores a human body determination result of the previous determination information in the image information captured by the imaging means, and the control means stores a current captured image obtained by newly capturing an image. 3. The method according to claim 1, wherein in the determination information, a third determination process of determining whether the object is the first object is performed based on at least a previous result of the human body determination. 4. Object detection device. 4. A display device for displaying the image information,
4. The object detection device according to claim 1, wherein the control unit controls the display unit to highlight the shape of the first object on the display. 5. The object detecting apparatus according to claim 1, wherein said storage means stores first object information corresponding to shape characteristics of a person or an animal. .