JP2000113374A - Device for extracting vehicle - Google Patents

Abstract

PROBLEM TO BE SOLVED: To effectively remove a shadow, and to surely extract only a vehicle by differentiating a multi-level image, discriminating it dense when the differential value is a prescribed value or more, and thin when the differential value is less than the prescribe value, and deciding the part discriminated dense as a vehicle, and deciding the part discriminated thin as a shadow. SOLUTION: Image data image pickup-up by an image inputting part 10 are converted through an A/D converter into a multi-level image, and stored in an image data storing part 11 constituted of an RAM. The multi-level image picked-up by a camera and stored in the image data storing part 11 extracted, and a differentiation processing is operated in a direction perpendicular to the traveling direction of the vehicle. Then, when the differential value is a prescribed value or more, the image is discriminated dense, and when the differential value is less than the prescribed value, the image is discriminated thin, and the part discriminated dense is decided as a vehicle, and the part discriminated thin is decided as a shadow. Then, the shadow is removed from the image data stored in the image data storing part 11 by a vehicle extracting part 12 constituted of a CPU as a center so that only the vehicle can be extracted.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a vehicle extracting device incorporated in, for example, an image type vehicle sensor, and more particularly to a vehicle extracting device capable of efficiently extracting only a vehicle by eliminating a shadow portion from captured image data. Regarding what you did.

[0002]

2. Description of the Related Art Conventionally, an image-type vehicle sensor in which a vehicle extraction device is incorporated captures an image of a vehicle traveling obliquely from above a road on which a vehicle (automobile) travels using a camera such as a CCD camera. The traffic flow parameters such as the cross-section traffic volume, speed, vehicle type, and congestion length are measured based on the image screen.

In the above-described traffic flow parameter measurement,
Although it is premised that a vehicle is extracted from the imaging screen, since the image type vehicle sensor is installed outdoors, it is necessary to perform a process of extracting only the vehicle by distinguishing the vehicle and the shadow from the imaging screen.

A conventional vehicle extracting device that extracts only a vehicle by distinguishing between the vehicle and the shadow, when a vehicle is extracted for each lane and one of them exists continuously in both lanes, one of the vehicles is extracted as a shadow of the vehicle. It is possible to determine and remove the shadow to extract only the vehicle (see the vehicle B and the shadow B of the vehicle A in FIG. 1), or a so-called difference from the difference between when the vehicle does not exist and when the vehicle exists. Vehicles are extracted by the processing.

[0005]

However, in the above-described conventional vehicle extraction device, the shadow of the vehicle can be removed by utilizing the movement of the vehicle along with the movement of the vehicle or by a difference process. A shadow that does not change according to the movement of the vehicle (see shadow B ′ in FIG. 1) is erroneously detected as a stopped vehicle. Also, the shadow fixed in this manner cannot create a reference screen serving as a background even in the difference processing method, and it is difficult to remove a shadow of a tree or the like, and the shadow of a tree or the like is extracted as a vehicle. There were drawbacks.

Accordingly, the present invention has been made to solve the above-mentioned drawbacks, and has as its object to provide a vehicle extracting apparatus capable of effectively removing shadows and reliably extracting only vehicles. To provide.

[0007]

In order to achieve the above object, a vehicle extracting means according to the present invention comprises: an image pickup means for picking up an image of a front surface of a vehicle; and an analog-to-digital conversion of a picked-up image into a multi-valued image. A converting means for converting, a differentiating means for differentiating the converted multi-valued image in a direction perpendicular to the traffic direction of the vehicle, densely when the differential value is equal to or more than a predetermined value, and sparse when the differential value is less than the predetermined value. And a determination means for determining a densely separated portion as a vehicle and a sparsely separated portion as a shadow.

[0008]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a perspective view of an image-type vehicle sensor to which a vehicle extraction device according to an embodiment of the present invention is applied, and FIG. 2 is a block diagram of the vehicle extraction device.

In FIG. 1, reference numeral 1 denotes a camera, such as a CCD camera, provided above a vehicle a traveling on a road R in the forward direction, and is provided on a gantry 2 erected on the road side.

The camera 1 is configured to be able to image a predetermined position in front of the camera 1. Therefore, when the vehicle A is located at the predetermined position, the front surface of the vehicle A is imaged by the camera 1.

In FIG. 2, reference numeral 10 denotes an image input section mainly constituted by the camera 1. Image data captured by the image input unit 10 is converted into a multi-valued image via an A / D converter (not shown), and is stored in an image data storage unit 11 including a RAM. ing.

The image data stored in the image data storage unit 11 is obtained by removing a shadow from the image data in a vehicle extraction unit 12 mainly constituted by a CPU constituting the vehicle extraction device of the present invention. Only those are extracted. Vehicle extraction unit 1
The vehicle information extracted in 2 is sent to a traffic signal control device (not shown) and used for control of the traffic signal and the like.

The image data storage section 11 corresponds to a RAM of a memory of a CPU constituting the vehicle extracting section 12, and a vehicle extracting function of the vehicle extracting section 12 is realized by arithmetic processing of the CPU.

The vehicle extracting unit 1 is based on the flowchart of FIG.
Before describing the vehicle control operation of FIG. 2, FIG.
The principle of vehicle extraction according to the present invention will be described with reference to FIG.

FIG. 4A schematically shows a state before an image differentiation process. FIG. 4B shows the value of the differentiation process. As is clear from FIGS. 4A and 4B, the front portion of the vehicle A having many horizontal stripes is
The differential value appears “dense”, and the differential value corresponding to the shadow b appears “sparse”. Therefore, it can be determined that the “dense” part of the differential value is a vehicle and the “sparse” part of the differential value is a shadow. The length of the “dense” portion (horizontal length in FIG. 4B) indicates the size of the vehicle, that is, the type of the vehicle such as large, medium, or small.

In the flowchart of FIG. 3, "dense"
Only the part is extracted. First, camera 1
The multi-valued image captured by the image data and stored in the image data storage unit 11 is extracted, and a differentiation process is performed in a direction perpendicular to the traveling direction of the vehicle A (Step 100, Step 1).
02. Hereinafter, the step is referred to as “S” in FIG. )
Further, a memory for storing a result determined as “dense” (a memory (not shown) provided in the vehicle extracting unit 12)
Is cleared, and an edge flag serving as a criterion for determining “sparse” and “dense” is prepared (S106, S10)
8).

In order to investigate "sparse" and "dense", a survey window of ± n (here, 3) dots is set and scanned in the vehicle traveling direction of the target pixel. As shown in FIG. 5, the survey window is A, A from the end (the upper end in FIG. 5).
B ... and the alphabet is assigned, where n =
Since it is 3, it is assigned to A to G. The difference between the assigned pixels, that is, AB = d1, BC = d2... Is calculated. Here, d1 to d6 are calculated (S1
08).

The calculated d1 to d6 are compared with a predetermined threshold value β for binarization. If the calculated threshold value is larger than the threshold value β (S110: YES), the signs (+,
−) Is recorded, and the edge flag is set to “+1” only when the previously recorded code is different from the current code (S110 to S120). When all the investigations for d1 and d2 are completed (Yes at S120), if the edge flag is equal to or more than a predetermined value (here, 2 or more), it is determined to be “dense” and stored in the memory prepared at S104 ( S
122 affirmative, S124). When the above process is performed over the entire image (Yes in S126), only the “dense” portion that hits the vehicle, that is, the “sparse” portion that hits the shadow, is stored in the memory.
Is deleted and only the vehicle is extracted.

The extracted vehicle is sent to a CPU for controlling a vehicle sensor (not shown), and the cross-sectional traffic, speed,
The type of vehicle or the length of traffic jam is measured.

[0020]

According to the present invention, there is provided a vehicle extracting apparatus, comprising: an image pickup unit for picking up an image of a front surface of a vehicle; a converting unit for converting a picked-up image from analog to digital to convert it to a multi-valued image; Differentiation processing means for differentiating in the direction perpendicular to the traffic direction of the vehicle, and separation means for densely separating when the differential value is equal to or more than a predetermined value, and sparsely when the differential value is less than the predetermined value,
Since it consists of a judgment means for judging a densely separated portion as a vehicle and judging a sparsely separated portion as a shadow, the shadow of a vehicle or a tree is efficiently removed, and the vehicle is reliably extracted. can do.

[Brief description of the drawings]

FIG. 1 is a perspective view of an image type vehicle sensor to which a vehicle extraction device according to an embodiment of the present invention is applied.

FIG. 2 is a block diagram of a vehicle extraction device.

FIG. 3 is a flowchart illustrating a vehicle extraction control operation.

FIG. 4 is an explanatory diagram showing the principle of vehicle extraction.

FIG. 5 is an explanatory diagram showing a survey range.

[Explanation of symbols]

 A vehicle b, b 'shadow 1 camera 2 gantry 10 image input unit 11 image data storage unit 12 vehicle extraction unit (vehicle extraction device)

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor Hisashi Kurosaki 1-13-8 Kamikizaki, Urawa-shi, Saitama Japan Signaling Corporation Yono Works F-term (reference) 5H180 AA01 CC04 DD04 EE07 5L096 AA06 BA04 EA35 GA02 GA17 LA04 9A001 BB02 BB04 EE02 EE05 GG13 HH21 HH24 HH25 JJ77 KK37 KZ16 KZ31 KZ56

Claims (1)

[Claims] 1. An image capturing means for capturing an image of a front surface of a vehicle, a converting means for converting a captured image from analog to digital to convert it to a multi-valued image, and a direction perpendicular to the traffic direction of the vehicle. A differential processing means for differentiating the differential value, a separating means for densely separating when the differential value is equal to or more than a predetermined value, and a separating means for sparsely separating when the differential value is less than the predetermined value. A vehicle extraction device comprising: a determination unit configured to determine a classified portion as a shadow;