JP2007114831A - Object detection device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an object detection device capable of simplifying and speeding up individual identification of an object. <P>SOLUTION: A laser radar device 7 detects the object inside a detection area (S101), obtains a reflected light amount data (S103), and acquires a direction and a distance from its own vehicle 101 to the object (S103). The radar device 7 distinguishes an attribute of the detected object as a high reflection body or a low reflection body on the basis of the acquired reflected light amount data (S104). A CCD camera 4 images the detection area (S105), and an image data generation part 5 generates image data (S106). An image processing part 6 sets a matching processing area on the image data from attribute information and object detection information obtained by the radar device 7, and performs pattern matching by a sampling pattern according to each the attribute. The image processing part 6 performs the individual identification of the object on the basis of a matching result thereof (S107). <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an apparatus for detecting an object, and more particularly to an object detection apparatus for determining an attribute of a detected object.

2. Description of the Related Art Conventionally, an image that is mounted on an automobile and that performs image processing on an image acquired by a radar device such as a laser radar and an imaging device such as a CCD camera as an object detection device that detects road conditions in front of the host vehicle or other vehicles. Many of them are equipped with processing devices (see, for example, Patent Document 1 and Patent Document 2). Most of such object detection devices detect each object with a radar device, and also detect position information of the object including a distance from the own vehicle to each object and an orientation of each object with respect to the own vehicle. Then, the object detection apparatus performs individual identification such as the shape and type of each object by mapping the position information onto the image data and analyzing the image data by the image processing apparatus.
JP 2003-252147 A JP 2003-217099 A

However, when the object detection information obtained by the radar device is mapped to image data and an object is individually identified, there are the following problems.
FIG. 13A is a schematic diagram showing a position state when a reflector of a preceding vehicle is irradiated with laser light, and FIG. 13B is a schematic diagram showing a position state of an object when a person is irradiated with laser light. .
FIG. 14A is a diagram showing the amount of reflected light of a high reflector (reflector = regressive reflector) in laser radar in azimuth direction, and FIG. 14B is the amount of reflected light of a low reflector (such as a person) in laser radar. (C) is a diagram in which (A) and (B) are superimposed.

  As shown in FIG. 13 (A), when the reflector (high reflector) 121A of the preceding vehicle 102 exists in the direction θA from the own vehicle 101, the reflected light amount has a peak value in the direction θA as shown in FIG. 14 (A). P2A is detected. Further, as shown in FIG. 13B, when a person 103A exists from the own vehicle 101 in the same direction θA, as shown in FIG. 14B, the peak value P3A is detected at the direction θA as shown in FIG. 14B.

  Similarly, as shown in FIG. 13 (A), when the reflector (high reflector) 121B of the preceding vehicle 102 exists in the direction θB from the own vehicle 101, the reflected light amount is in the direction θB as shown in FIG. 14 (A). Thus, the peak value P2B is detected. Further, as shown in FIG. 13B, when a person 103B exists in the same direction θB from the own vehicle 101, the peak value P3B is detected at the direction θB as shown in FIG. 14B.

  In this way, when the reflector 121A and the person 103A of the preceding vehicle 102 exist in the same direction θA, and the reflector 121B and the person 103B of the preceding vehicle 102 exist in the same direction θB, as shown in FIG. The peak generation directions of the reflected light quantity characteristics are the same.

Further, the laser beam has a strong straightness and the beam width is difficult to spread, but it does not spread completely. For this reason, if the beam width at the position where the reflectors 121A and 121B and the persons 103A and 103B are present is wider than the distance between the reflectors 121A and 121B and the distance between the persons 103A and 103B, the laser beam is reflected by them. A certain amount of reflected light is obtained.
As a result, as shown in FIG. 14C, the reflected light amount characteristic curve is substantially the same regardless of whether the level of the reflected light amount is different between the reflector (high reflector) and the human (low reflector). Will be obtained.

  For this reason, when the conventional object detection device detects the peak of the reflected light amount, it calculates the azimuth θ corresponding to this, and adapts various sampling patterns to the region corresponding to the azimuth θ of the image data. The individual identification of the object was performed.

  However, in this method, since it is not known whether the object corresponding to the peak value is a reflector of a preceding vehicle or a person, every time a peak is detected, all stored sampling patterns are adapted and optimized. A sampling pattern must be extracted. For this reason, the processing becomes complicated, and high-speed object identification cannot be performed.

  Therefore, an object of the present invention is to provide an object detection apparatus that can simplify and speed up individual identification of an object.

  The object detection device of the present invention discriminates an attribute of an object based on position information acquisition means for acquiring the reflection intensity due to reflection of the object in the detection area and at least the distance of the object, and the acquired distance and reflection intensity. And an attribute discrimination means.

  In this configuration, the position information acquisition unit receives the reflection signal from the object existing in the detection region, detects at least the distance to the object, and acquires the reflection signal intensity in association with this. The attribute discrimination means discriminates the attribute of the object from the obtained distance of the object and the reflected signal intensity. More specifically, a plurality of intensity regions are set by providing a discrimination threshold value for reflected signal intensity.

  The attribute discrimination means discriminates which intensity region the obtained reflected signal intensity belongs to with respect to the detected distance. Then, the attribute discrimination means outputs the attribute of the object, specifically, whether it is a high reflector or a low reflector based on this discrimination result.

  Further, the attribute discrimination means of the object detection apparatus of the present invention sets a discrimination threshold using at least one of the light intensity data of the reflection intensity characteristic, the light intensity data by distance, the light intensity variation data, and the distance variation data, and the object attribute It is characterized by distinguishing.

  In this configuration, the attribute discrimination means specifically selects and sets one of the following four threshold setting methods as the discrimination threshold.

(1) The threshold value is determined with a constant amount of reflected light that does not depend on the distance by utilizing the fact that the reflection intensity differs by two digits or more between the high reflector and the low reflector.
(2) With reference to the fact that the reflection intensity is attenuated as the distance becomes longer for both the high reflector and the low reflector, the threshold of the amount of reflected light is determined while changing according to the distance.
(3) Utilizing the fact that the variation in the reflection intensity of the low reflector is larger than that of the high reflector, a specific amount of reflected light amount variation is set as the threshold value.
(4) Utilizing the fact that the variation in the distance at which the low reflector is detected is greater than that of the high reflector, the detection distance variation of a specific amount is set as a threshold value.
By these methods, a high reflector and a low reflector are clearly distinguished.

  In addition, the object detection device of the present invention performs imaging processing for capturing an image of a detection region and generating image data, and performs image processing different for each attribute in association with the detected object corresponding to the image data, thereby identifying an individual object. And an individual identification means for performing the identification.

  In this configuration, the imaging unit acquires image data of a detection area where the position information acquisition unit detects an object. The individual identifying means maps the position of the detected object on the image data, and sets a different image processing region and sampling pattern for each corresponding attribute. The individual identifying means identifies the pattern of the object in the set image area using a plurality of individual sampling patterns prepared in advance for each image processing area. And if a pattern is identified, based on this pattern, the kind of object, for example, a car, a delineator, a person, an object, etc. will be output.

  In addition, the individual identifying means of the object detecting device of the present invention is characterized in that, in particular, the regressive reflector is identified from other objects. In this configuration, specifically, the retroreflector, which is a reflector provided at the rear part of the preceding vehicle, is identified from other objects such as a person.

  In addition, the object detection apparatus of the present invention is characterized in that the position information acquisition means is realized by a laser radar that irradiates at least a detection region with laser light and uses the reflected light. In this configuration, specifically, a laser radar is used as at least a position information acquisition device.

  According to the present invention, the individual identification processing in the subsequent stage is easily and rapidly performed by first classifying the objects in predetermined attribute units from the distance and the amount of reflected light obtained by the position information acquisition means such as a laser radar. can do. Then, after roughly classifying the object based on the attribute in this way, the individual identification is performed in detail, so that each process is simple and high speed. Therefore, the object identification can be performed more easily and faster than the conventional individual identification method. Identification can be made.

An object detection apparatus according to a first embodiment of the present invention will be described with reference to FIGS.
FIG. 1A is a block diagram showing the main part of the object detection device of the present embodiment, and FIG. 1B is a conceptual diagram showing the detection state of an automobile equipped with the object detection device of the present embodiment.
The object detection apparatus of this embodiment includes a radar antenna 1, a radar signal processing unit 2, an attribute detection unit 3, a CCD (stereo) camera 4, an image data generation unit 5, and an image processing unit 6.

  The radar antenna 1 is installed near the lower end of the front bumper of the automobile (own vehicle) 101, transmits a laser beam in front of the own vehicle 101, and receives reflected light formed by reflecting the laser beam to an object in the detection area. To do. The radar antenna 1 is installed so as to be rotatable in the horizontal direction with respect to the own vehicle 101, and scans the laser beam in the horizontal direction by rotating the radar antenna 1 within a predetermined azimuth angle range.

  The radar signal processing unit 2 includes a laser beam control unit and a reception signal analysis unit, and appropriately applies a laser control signal generated by the laser beam control unit to the radar antenna 1 to thereby transmit a laser beam transmitted in a predetermined direction. Form. When the radar signal processing unit 2 receives reflected light from the radar antenna 1, the radar signal processing unit 2 detects the intensity of the received signal, that is, the amount of reflected light, and detects the azimuth direction. Further, the radar signal processing unit 2 measures the distance between the own vehicle 101 and the detected object from the reception timing of the corresponding reception signal and the transmission timing of the transmission laser beam. Thereby, an object within a predetermined angular range in the horizontal direction is detected.

The attribute detection unit 3 uses the object detection result output from the radar signal processing unit 2 to detect the attribute of the object detected by a method described later. Here, the attribute is set based on the amount of reflected light, and includes, for example, a high reflector having a large amount of reflected light such as a reflector and a low reflector having a small amount of reflected light such as a person.
The radar signal processing unit 2 and the attribute detection unit 3 are integrated as a laser radar device 7.

The CCD camera 4 is installed in the vicinity of the upper end portion of the windshield of the host vehicle 101 and images the road situation in front of the host vehicle 101.
The image data generation unit 5 converts the video captured by the CCD camera 4 into image data at every predetermined timing, and gives the image data to the image processing unit 6.
The image processing unit 6 performs individual identification of each object by a method described later based on the object detection information, direction, distance, and attribute given from the laser radar device 7.
The image data generation unit 5 and the image processing unit 6 are integrated as an image processing device 8.

Next, the process of the object detection device of the present embodiment will be specifically described with reference to FIGS.
FIG. 2 is a flowchart showing a main operation flow of the object detection apparatus of the present embodiment.
FIG. 3 is a flowchart showing a processing flow of attribute discrimination. FIG. 3A shows a case where the attribute is discriminated into two types, that is, a high reflector and a low reflector, and FIG. The case where it discriminate | determines in three types, a body and a low reflector is shown. The high reflector described here is, for example, a reflector provided in an automobile or a delineator, and the medium reflector and the low reflector are, for example, a person or an object (obstacle).
FIG. 4 is a diagram for explaining the concept of setting the reflected light amount threshold TH for attribute discrimination. FIG. 4A shows a threshold TH0 for discriminating the attribute into two types, a high reflector and a low reflector, and FIG. The first threshold value TH1 and the second threshold value TH2 are shown for discriminating into three types of high reflector, medium reflector and low reflector. In FIG. 4, the thick solid line indicates the reflected light amount characteristic curve of the high reflector, the solid solid line indicates the reflected light amount characteristic curve of the medium reflector, and the thin solid line indicates the reflected light amount characteristic curve of the low reflector.

  As described above, the radar signal processing unit 2 of the laser radar device 7 transmits a laser beam while scanning the front of the host vehicle 101 in the horizontal direction and receives reflected light from the object, thereby Detection is performed (S101). At this time, the object detection data includes the direction of the detected object and the reception timing (reception time). At the same time, the radar signal processing unit 2 detects the intensity of the reflected light, that is, the amount of reflected light, and gives it to the object detection data (S102). The radar signal processing unit 2 calculates the distance of the detected object from the obtained object detection information, and calculates object position data including the azimuth and the distance (S103). As is known, the distance is calculated from the time difference between the transmission timing and the reception timing.

The attribute detection unit 3 of the laser radar device 7 performs attribute discrimination using the object detection data and the object position data (S104).
(1) When discriminating attributes into two types, a high reflector and a low reflector (see FIG. 3A)
The attribute detection unit 3 reads the reflected light amount data and compares it with a preset reflected light amount threshold TH0. Here, as shown in FIG. 4A, the reflected light amount threshold TH0 is such that the light amount region that can be taken by the high reflector and the light amount region that can be taken by the low reflector are overlapped regardless of the distance. It is set by using the incompatibility. That is, an appropriate amount of reflected light between the minimum value of the reflected light amount of the high reflector and the maximum value of the reflected light amount of the low reflector, for example, the minimum value of the reflected light amount of the high reflector and the reflected light amount of the low reflector An intermediate value from the maximum value is set to the reflected light amount threshold TH0.

  Then, if the detected reflected light amount data is larger than the threshold value TH0, the attribute detection unit 3 determines the detected object as a high reflector (S401 → S402). On the other hand, if the detected amount of reflected light is smaller than the threshold value TH0, the attribute detection unit 3 determines that the detected object is a low reflector (S401 → S403).

(2) When discriminating attributes into three types: high reflector, medium reflector, and low reflector (see FIG. 3B)
The attribute detection unit 3 reads the reflected light amount data and compares it with a preset reflected light amount first threshold TH1. Here, as shown in FIG. 4B, the reflected light amount first threshold value TH1 includes a light amount region that can be taken by the high reflector and a light amount region that can be taken by the middle reflector and the low reflector. It is set using the fact that it does not overlap regardless of the distance. That is, an appropriate amount of reflected light between the minimum value of the reflected light amount of the high reflector and the maximum value of the reflected light amount of the medium reflector and the low reflector, for example, the minimum value of the reflected light amount of these high reflectors and the medium reflector Further, an intermediate value between the maximum value of the reflected light amount of the low reflector and the reflected light amount first threshold value TH1 is set.

  And the attribute detection part 3 will discriminate | determine a detection object as a high reflector, if the detected reflected light amount data is larger than 1st threshold value TH1 (S404-> S405). On the other hand, if the detected amount of reflected light is smaller than the first threshold TH1, the attribute detection unit 3 determines that the detected object is not a high reflector, and performs the next determination process (S404 → S406).

  Next, the attribute detection unit 3 compares the reflected light amount data with a preset reflected light amount second threshold TH2. Here, as shown in FIG. 4B, the reflected light amount second threshold TH2 is related to the distance between the light amount region that can be taken by the intermediate reflector and the light amount region that can be taken by the low reflector. It is set using the fact that there is no overlap. That is, an appropriate amount of reflected light between the minimum value of the reflected light amount of the medium reflector and the maximum value of the reflected light amount of the low reflector, for example, the minimum value of the reflected light amount of these medium reflectors and the reflected light amount of the low reflector An intermediate value from the maximum value is set to the reflected light amount second threshold TH2.

  If the reflected light amount data is larger than the second threshold value TH2, the attribute detection unit 3 determines that the detected object is a medium reflector (S406 → S407). On the other hand, if the detected amount of reflected light is smaller than the second threshold TH2, the attribute detection unit 3 determines that the detected object is a low reflector (S406 → S408).

  Such attribute discrimination processing is performed for all of the detected objects. When the attribute is determined, the laser radar device 7 associates the object detection data, the object position data, and the attribute data with the image processing device. Give to 8.

  In parallel with the operation of the laser radar device 7, the CCD camera 4 images the front of the host vehicle 101 (S 105) and outputs it to the image processing device 8. The image data generation unit 5 of the image processing device 8 sequentially forms image data at a predetermined timing and gives the image data to the image processing unit 6 (S106). At this time, image data generation timing, that is, imaging timing is attached to the image data.

  The image processing unit 6 synchronizes the time based on the object detection data, object position data, and attribute data given from the laser radar device 7 and the image data given from the image data generation unit 5. Individual identification is performed (S107).

  Specifically, first, the image processing unit 6 maps the object detection points on the image data based on the object position data. In addition, about the individual identification process started from the following mapping processes, the case where it distinguishes into two types, a high reflector and a low reflector, is demonstrated.

(1) When the attribute of the detected object is a high reflector FIG. 5 is a flowchart for explaining individual identification processing when the attribute of the detected object is a high reflector.
FIG. 6 is a diagram for explaining individual identification processing when the attribute of the detected object is a high reflector. 6A shows a matching area setting state, FIG. 6B shows a state where only the matching area is extracted, and FIGS. 6C and 6D show sampling patterns used for matching.

  The image processing unit 6 sets a matching processing region from the corresponding attribute data for the mapped object detection point (S701). Specifically, when the image processing unit 6 identifies that the detected object is a high reflector, the image processing unit 6 sets a matching processing region 12 including two object detection points that appear at predetermined intervals.

  At this time, since the high reflector is a reflector installed at the rear part of the preceding vehicle or a delineator installed on the side of the road or on the road surface, the matching area is set to a rectangular shape according to the shape of the automobile. Specifically, the matching region 12 is set to have a width so that the two object detection points are positioned at both ends in the width direction, and set to a height that is set separately in advance. Here, the set height is set from the calculated value obtained by, for example, calculating in advance a size of a general automobile that can be seen forward from the own vehicle based on the distance from the own vehicle.

  Next, the image processing unit 6 performs image processing such as outline extraction based on contrast on the extracted image data of the matching region 12 to detect the shape of the object including the object detection point (S702).

  The image processing unit 6 stores sampling patterns 201 to 203 as shown in FIGS. 6C and 6D in the memory in advance, and sequentially reads these sampling patterns 201 to 203 (S703). As shown in FIGS. 6C and 6D, the sampling patterns 201 to 203 for the high-reflector object are set in accordance with the shape of the automobile in the same manner as the matching area 12, and are previously set for various automobiles. It is set in the back shape (the shape seen from the back).

  Note that when the sampling pattern is read and used, the image processing unit 6 enlarges or reduces the stored sampling pattern based on the distance information obtained for the object. For example, if a sampling pattern with an image size of m dots × n dots is used and the detection distance is L with respect to the reference distance (the distance used when the sampling pattern is set) L0, the image used for pattern matching The size is converted into m × (L0 / L) dots × n × (L0 / L) dots.

  Next, the image processing unit 6 performs a pattern matching process using the detected object shape and each of the sampling patterns 201 to 203 to search for a similar pattern (S704). The image processing unit 6 verifies the similarity between the detected object shape and the sampling pattern by a known method, and adapts this sampling pattern if there is a pattern similar to a preset reference or more. Then, the image processing unit 6 reads the object type corresponding to the similar pattern and limits the object type. For example, in the case shown in FIG. 6, since the sampling pattern is the shape of an automobile, if there is a similar pattern, it is determined that the detected object is an automobile corresponding to the similar pattern (S705 → S706). Here, if the sampling patterns of various automobiles, for example, ordinary automobiles, light trucks, large trucks, etc. are individually set, the vehicle type can also be detected.

  Further, the image processing unit 6 detects the size of the object using the selected sampling pattern. For example, if the determined similar pattern is m1 dots × n1 dots in terms of the number of pixels and the distance of the object is L, if the focal length of the laser radar is f and the pixel pitch distance of the image data is p, then M ( = M1 * p * L1 / f) * N (= n1 * p * L1 / f).

  By the way, when there are a plurality of similar sampling patterns, the image processing unit 6 may select the sampling pattern having the strongest similarity from the similar sampling patterns.

On the other hand, if the similarity between all the sampling patterns and the shape based on the detected object does not reach the predetermined reference, the image processing unit 6 determines that the detected high reflector is not a vehicle reflector (S705). → S707). For example, when it is detected that high reflectors that are not determined as vehicle reflectors continue in a predetermined direction, they are determined to be delineators.
Then, the image processing unit 6 outputs the detection result of the object determined in this way to a higher system on which the object detection device is mounted (S708).

(2) When the attribute of the detected object is a low reflector FIG. 7 is a flowchart for explaining individual identification processing when the attribute of the detected object is a low reflector.
FIG. 8 is a diagram for explaining individual identification processing when the attribute of the detected object is a low reflector. 8A shows a matching area setting state, FIG. 8B shows a state where only the matching area is extracted, and FIGS. 8C and 9D show sampling patterns used for matching.

  The image processing unit 6 sets a matching processing area from the corresponding attribute data for the mapped object detection point (S711). Specifically, when the image processing unit 6 identifies that the detected object is a low-reflector, the image processing unit 6 sets matching processing regions 12A and 12B including each object detection point for each object detection point.

  At this time, since the low reflector is a person who stops or moves on the road, or is an object (obstacle, luggage, etc.) placed on the road, the matching area is determined according to the shape of the person. Set. Specifically, the matching areas 12A and 12B are set to rectangular areas having a width and a height in accordance with the outline of the person with the object detection point as the center in the width and height directions. Here, the set width and height are set from the calculated values obtained by, for example, calculating in advance the dimensions of a general person who can see forward from the own vehicle based on the distance from the own vehicle.

  Next, the image processing unit 6 performs image processing such as outline extraction by contrast on the extracted image data of the matching regions 12A and 12B, and detects the shape of the object including the object detection point (S712).

  Further, the image processing unit 6 stores sampling patterns 301 to 303 as shown in FIGS. 8C and 8D in a memory in advance, and sequentially reads these sampling patterns 301 to 303 (S713). As shown in FIGS. 8C and 8D, the sampling patterns 301 to 303 for the low-reflector object are set in accordance with the shape of the person in the same way as the matching areas 12A and 12B, and various people are previously set. It is set by the back shape (shape seen from the back) of the action pattern.

  Also in this case, when the image processing unit 6 reads and uses the sampling pattern, the image processing unit 6 enlarges or reduces the stored sampling pattern based on the distance information obtained for the object.

  Next, the image processing unit 6 performs a pattern matching process using the detected object shape and each of the sampling patterns 301 to 303 to search for a similar pattern (S714). Then, the image processing unit 6 verifies the similarity between the detected object shape and the sampling pattern by a known method, and adapts this sampling pattern if there is a pattern that is more than a preset reference. Then, the image processing unit 6 reads the object type corresponding to the similar pattern and limits the object type. For example, in the case shown in FIG. 8, since the sampling pattern has a human shape, if a similar pattern exists, it is determined that the detected object is a person corresponding to this similar pattern (S715 → S716). Here, when there are a plurality of similar sampling patterns, the most similar sampling pattern is selected from these.

  On the other hand, if the similarity between all the sampling patterns 301 to 303 and the shape based on the detected object does not reach the predetermined reference, the image processing unit 6 determines that the detected low reflector is not a person ( S715 → S717). For example, a low reflector that is not determined as a person in this way is determined to be an obstacle present on a road, for example.

  Then, the image processing unit 6 outputs the detection result of the object determined in this way to a higher system on which the object detection device is mounted (S718).

  As for the discrimination of such a person, as described above, the middle reflector and the low reflector are discriminated. For example, an adult sampling pattern is used as the middle reflector, and a child sampling pattern is used as the low reflector. In this way, the individual identification process may be performed.

  By performing such processing, according to the attribute of the object (in this embodiment, the reflection intensity), the object is classified into a predetermined type, and a unique matching process is performed for each attribute, so that the individual object Identify. As a result, the matching processing area can be uniquely set for each attribute, and an individual matching sample pattern can be read for each attribute, so that the matching processing can be simplified and speeded up.

Next, an object detection apparatus according to a second embodiment will be described with reference to FIGS.
In the first embodiment, an example in which a threshold value that does not depend on the distance is used for attribute determination has been described. However, the object detection device of the present embodiment uses a threshold value that is set according to the distance. Is.
FIG. 9 is a flowchart showing a process flow of attribute discrimination. FIG. 9A shows a case where the attribute is discriminated into two types, that is, a high reflector and a low reflector, and FIG. The case where it discriminate | determines in three types, a body and a low reflector is shown.
FIG. 10 is a diagram for explaining the concept of setting the reflected light amount threshold THn for attribute discrimination. FIG. 10A shows the threshold THn0 for discriminating into two types of attribute, high reflector, and low reflector, and FIG. The first threshold value THn1 and the second threshold value THn2 are shown for discriminating into three types of high reflectors, medium reflectors, and low reflectors. In FIG. 10, the thick solid line shows the reflected light amount characteristic curve of the high reflector, the solid solid line shows the reflected light amount characteristic curve of the medium reflector, and the thin solid line shows the reflected light amount characteristic curve of the low reflector.

(1) When discriminating the attribute into two types, a high reflector and a low reflector (see FIG. 9A)
Upon receiving the object detection data, the attribute detection unit 3 reads the distance and the reflected light amount data, and compares it with the reflected light amount threshold THn0 at the corresponding distance (S411 → S412 → S413). Here, as shown in FIG. 10A, the reflected light amount threshold THn0 uses that the reflected light amount of the high reflector and the reflected light amount of the low reflector are very far apart regardless of the distance selected. To be preset. That is, an appropriate reflected light amount between the reflected light amount of the high reflector and the reflected light amount of the low reflector at each distance, for example, an intermediate value between the reflected light amount of the high reflector and the reflected light amount of the low reflector at the corresponding distance Is set to the reflected light amount threshold THn0.

  If the detected reflected light amount data is larger than the threshold THn0 at the corresponding distance, the attribute detection unit 3 determines that the detected object is a high reflector (S413 → S414). On the other hand, if the detected amount of reflected light is smaller than the threshold value THn0 at the corresponding distance, the attribute detection unit 3 determines the detected object as a low reflector (S413 → S415).

(2) When distinguishing three types of attributes: high reflector, medium reflector, and low reflector (see FIG. 9B)
Upon receiving the object detection data, the attribute detection unit 3 reads the distance and the reflected light amount data, and compares it with the reflected light amount first threshold THn1 at the corresponding distance (S411 → S412 → S416). Here, as shown in FIG. 10 (B), the reflected light amount first threshold value TH1 is very different from the reflected light amount of the high reflector and the reflected light amount of the middle reflector regardless of the distance selected. Is set in advance using. That is, an appropriate amount of reflected light between the amount of reflected light of the high reflector and the amount of reflected light of the middle reflector at each distance, for example, the minimum value of the reflected light amount of these high reflectors and the maximum value of the reflected light amount of the middle reflector Is set to the reflected light amount first threshold value THn1.

  If the detected reflected light amount data is larger than the first threshold value THn1, the attribute detection unit 3 determines that the detected object is a high reflector (S416 → S414). On the other hand, if the detected amount of reflected light is smaller than the first threshold THn1, the attribute detection unit 3 determines that the detected object is not a high reflector, and performs the next determination process (S416 → S417).

  Next, the attribute detection unit 3 compares the reflected light amount data with a reflected light amount second threshold THn2 set in advance. Here, as shown in FIG. 10B, the reflected light amount second threshold value THn2 is very different from the distance between the reflected light amount of the medium reflector and the reflected light amount of the low reflector. Is set in advance using. That is, an appropriate amount of reflected light between the amount of reflected light of the medium reflector and the amount of reflected light of the low reflector at each distance, for example, an intermediate value between the amount of reflected light of the medium reflector and the amount of reflected light of the low reflector is reflected. The light amount second threshold value THn2 is set.

  If the reflected light amount data is greater than the second threshold value THn2 at the corresponding distance, the attribute detection unit 3 determines that the detected object is a middle reflector (S417 → S418). On the other hand, if the detected amount of reflected light is smaller than the second threshold THn2 at the corresponding distance, the attribute detection unit 3 determines that the detected object is a low reflector (S417 → S419).

  Even with such a processing method, the same effects as those of the first embodiment can be obtained. Furthermore, by using the processing method of the present embodiment, the lowest value of the reflected light amount of the high reflector (the amount of reflected light from the farthest possible observation) is the highest value of the reflected light amount of the low reflector and the middle reflector (observed). Even if it is smaller than the nearest reflected light amount to be obtained, it is possible to reliably discriminate between a high reflector and a medium reflector or a low reflector.

Next, an object detection apparatus according to a third embodiment will be described with reference to FIGS.
In the first and second embodiments, the example in which the light quantity is used as a reference for the attribute determination is shown. However, the object detection apparatus according to the present embodiment has a threshold value set according to the variation in the distance and the light quantity. Is used. As shown in the first embodiment, the attribute determination flow shown in the present embodiment is performed after the attribute determination is performed on the high reflector and the non-high reflector, and then the non-high reflector is changed from the medium reflector to the low reflector. The case where the attribute is discriminated from the reflector is shown.
FIG. 11 is a flowchart showing a process flow of attribute discrimination. FIG. 11A shows a case of discriminating into two types of an attribute, a high reflector, and a low reflector, and FIG. The case where it discriminate | determines in three types, a body and a low reflector is shown.
12A is a diagram for explaining the setting concept of the reflected light amount variation threshold value σL0 for attribute discrimination, and FIG. 12B is a diagram for explaining the setting concept of the distance variation threshold value σD0 for attribute discrimination. In FIG. 12, a solid line indicates a variation characteristic curve of the medium reflector, and a thin solid line indicates a variation characteristic curve of the low reflector.

  The medium reflector and the low reflector are often people or objects as shown in the above example. For this reason, there is a tendency that the difference in the amount of reflected light between the medium reflector and the low reflector is not large, or that the amount of reflected light obtained for each measurement on the same object varies greatly. Furthermore, since the amount of reflected light is low, the calculated distance also varies. However, when comparing the medium reflector and the low reflector, the low reflector tends to increase both in the amount of reflected light and in the distance, so in this embodiment, using this characteristic, Discriminate between medium reflector and low reflector.

(1) When discriminating between a medium reflector and a low reflector based on variations in the amount of reflected light (see FIG. 11A)
When acquiring the reflected light amount data, the attribute detecting unit 3 confirms the azimuth direction and accumulates the reflected light amount data in the same azimuth direction for a predetermined time (for a plurality of scans) (S421). The attribute detection unit 3 calculates the reflected light amount variation data σL using these reflected light amount data, and compares it with a preset reflected light amount variation threshold σL0 (S422 → S423). Here, as shown in FIG. 12A, the reflected light amount variation threshold σL0 is a distance between the data region where the reflected light amount variation of the medium reflector can take and the data region where the reflected light amount variation of the low reflector can take. Regardless of whether they overlap, set. That is, an appropriate reflected light amount variation between the minimum value of the reflected light amount variation of the low reflector and the maximum reflected light amount variation of the medium reflector, for example, the minimum value of the reflected light amount variation of the low reflector and the medium reflector An intermediate value with the maximum value of the reflected light amount variation is set to the reflected light amount variation threshold σL0.

  If the detected reflected light amount variation data σL is smaller than the threshold σL0, the attribute detection unit 3 determines that the detected object is a medium reflector (S423 → S424). On the other hand, if the detected reflected light amount variation σL is larger than the threshold σL0, the attribute detection unit 3 determines the detected object as a low reflector (S423 → S425).

(2) When discriminating between the medium reflector and the low reflector based on the distance variation (see FIG. 11B)
When obtaining the distance data of the object detection data, the attribute detection unit 3 confirms the azimuth direction and accumulates the distance data in the same azimuth direction over a predetermined time (for a plurality of scans) (S431). The attribute detection unit 3 calculates the distance variation data σD using these distance data, and compares it with a preset distance variation threshold σD0 (S432 → S433). Here, as shown in FIG. 12 (B), the distance variation threshold σD0 is such that the data region where the distance variation due to the middle reflector can take and the data region where the distance variation due to the low reflector can take place overlap regardless of the distance. Set by using what does not fit. That is, an appropriate distance variation between the minimum value of the distance variation of the low reflector and the maximum value of the distance variation of the medium reflector, for example, the minimum value of the distance variation of the low reflector and the distance variation of the medium reflector. An intermediate value with respect to the maximum value is set to the distance variation threshold σD0.

  If the detected distance variation data σD is smaller than the threshold σD0, the attribute detection unit 3 determines that the detected object is a middle reflector (S433 → S434). On the other hand, if the detected distance variation σD is larger than the threshold σD0, the attribute detection unit 3 determines the detected object as a low reflector (S433 → S435).

  Even if such processing is performed, the same effects as those of the above-described embodiments can be obtained. Furthermore, by performing the processing of the present embodiment, it is possible to reliably discriminate between attributes that are difficult to discriminate with only the amount of reflected light.

In the present embodiment, an example is shown in which the medium reflector and the low reflector are discriminated based on the variation characteristics. However, even when the high reflector is included, the attribute discrimination can be performed similarly. That is, it is possible to perform attribute discrimination of the high reflector, the medium reflector, and the low reflector only by the variation characteristic.
In the above description, the case where the radar signal processing unit 2 and the attribute detection unit 3 are incorporated in the laser radar device 7 has been described. However, the function of the image processing device 8 may be incorporated in the laser radar device 7. On the contrary, the function of the laser radar device 7 may be incorporated in the image processing device 8.
Furthermore, only the attribute detection unit 3 of the laser radar device 7 may be incorporated in the image processing device 8. That is, the laser radar device 7 may only detect an object (detect the position and distance of the object), and the above-described attribute detection of the detected object may be performed by the image processing device 8.

In the above description, the laser radar is shown as an example of the radar apparatus. However, a radar apparatus of another method (pulse Doppler, FMCW, etc.) may be used, and further, from an object existing in the detection area. A device that performs triangulation using the reflected light may be used.
In the above description, an example is shown in which the type (individual identification) and shape of the detected object are identified. By knowing these, it is possible to count each type of object. For example, it is possible to recognize how many cars are traveling in front of the vehicle and how many people are present.

It is a block diagram which shows the principal part of the object detection apparatus of 1st Embodiment, and a conceptual diagram which shows the detection state of the motor vehicle carrying this object detection apparatus. It is a flowchart which shows the main operation | movement flow of the object detection apparatus of 1st Embodiment. It is a flowchart which shows the processing flow of attribute discrimination. It is a figure explaining the setting concept of the reflected light amount threshold value TH of attribute discrimination. It is a flowchart explaining the individual identification process in case the attribute of a detected object is a high reflector. It is a figure explaining the individual identification process in case the attribute of a detection object is a high reflector. It is a flowchart explaining the individual identification process in case the attribute of a detection object is a low reflector. It is a figure explaining the individual identification process in case the attribute of a detection object is a low reflector. It is a flowchart which shows the processing flow of the attribute discrimination | determination based on 2nd Embodiment. It is a figure explaining the setting concept of the reflected light amount threshold value THn of attribute discrimination. It is a flowchart which shows the processing flow of the attribute discrimination | determination concerning 3rd Embodiment. It is a figure explaining the setting concept of the reflected light amount variation threshold value (sigma) L0 of attribute discrimination, and the setting concept of the distance variation threshold value (sigma) D0 of attribute discrimination. It is a schematic diagram which shows the position state in case a laser beam is irradiated to the reflector of another vehicle, and a schematic diagram which shows the position state of the object in case a person is irradiated with a laser beam. FIG. 5 is a diagram showing the amount of reflected light of a high reflector (reflector = regressive reflector) in a laser radar in an azimuth direction, and a diagram showing the amount of reflected light of a low reflector (such as a person) in a laser radar in an azimuth direction. .

Explanation of symbols

1-radar antenna, 2-radar signal processor, 3-attribute detector, 4-CCD camera, 5-image data generator, 6-image processor, 7-laser radar device, 8-image processor

Claims (5)

Position information acquisition means for acquiring a reflection intensity by reflection of an object in the detection region and at least a distance of the object;
Attribute determining means for determining an attribute of the object based on the acquired distance and reflection intensity;
An object detection apparatus comprising:   The attribute determination unit sets the determination threshold using at least one of the light amount data of the reflection intensity characteristic, the light amount data by distance, the light amount variation data, and the distance variation data, and determines the attribute of the object. The object detection apparatus described. Imaging means for imaging the detection area and generating image data;
An individual identification unit that performs individual image processing for each attribute in association with the detected object corresponding to the image data, and performs individual identification of the object;
The object detection apparatus according to claim 1 or 2, further comprising:   The object detection apparatus according to claim 3, wherein the individual identification unit identifies a regressive reflector from other objects.   The object detection apparatus according to claim 1, wherein at least the position information acquisition unit is realized by a laser radar that irradiates a laser beam in the detection region and uses the reflected light.

Images