JP2965733B2 - Image-based vehicle sensing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method for processing a video signal obtained by capturing an image of a road and detecting a vehicle traveling on the road based on a change in luminance level corresponding to the same road position caused by the vehicle passing through the road. The present invention relates to a vehicle sensing method, and more particularly to a method for eliminating erroneous sensing of a traveling vehicle based on a shadow on a road surface.

[0002]

2. Description of the Related Art A method and apparatus have been put to practical use in which a road is imaged by an image pickup means in order to grasp a traffic situation, and an obtained video signal is processed to detect a vehicle passing through the road. Such an apparatus has been proposed by the present inventors as Japanese Utility Model Application No. 2-31892. In the image-based vehicle sensing method, an image-based vehicle sensing device as shown in FIG. 5 is generally used. This image-type vehicle sensing device includes a schematic image capturing unit 10, a luminance data conversion unit 20, an arithmetic processing unit 30, and a control unit 40. The image pickup means 10 is, for example, a CCD camera, and picks up an image of a road detection area as shown in FIG.
The video signal of the system is transmitted to the luminance data conversion unit 20. Using this video signal, the position of a plurality of measurement points (P ₁ , P ₂ ,...) Set on a straight line (detection line, L ₁ ) assumed in the cross direction of the road in the image of FIG. The presence of the vehicle on the detection line (L ₁ ) at a certain time is detected based on the difference between the luminance data groups obtained as a set of the luminance levels in the luminance signal corresponding to the luminance signal group at predetermined time intervals. . That is, as shown in FIG. 7, a luminance data group (reference road surface level data, D
r) and a vehicle is detected by detecting a difference between a luminance data group (vehicle detection level data, Di) when a vehicle is present on the detection line and the luminance level is largely changed along the detection line. . If a plurality of detection lines are set, the speed of the vehicle can be obtained based on the vehicle arrival time at both.

The luminance data converter 20 includes a synchronization signal extraction circuit 21 for obtaining a horizontal synchronization signal and a vertical synchronization signal from a video signal. Based on these synchronization signals, a current search position on a search line is digitized and corresponding horizontal coordinates are obtained. A horizontal address counter 22 and a vertical address counter 23 for obtaining vertical coordinates, a digital comparator 24 for comparing these outputs with coordinates sequentially designated by a microprocessor 41 (described later) and generating a capture signal at the moment of coincidence, A video A / D conversion circuit 25 quantifies the luminance of the video signal at the corresponding moment in accordance with the fetched signal, obtains luminance data, and outputs it. The input unit has a video amplifier 27 for amplifying the transmitted video signal to an appropriate level,
Following this, a clamp level fixing circuit 28 for normalizing the video signal to a constant amplitude prior to the A / D conversion is also provided. The arithmetic processing unit 30 is a circuit for receiving the luminance data from the video A / D conversion circuit 25 and averaging a plurality of luminance data and calculating a correlation. It uses a processor 31 and also has a storage unit necessary for storing data.

A control unit 40 controls the above-mentioned units, receives data from the arithmetic processing unit 30 and performs further arithmetic processing on the data to detect the presence of a vehicle, calculate the speed of the vehicle, and perform the following processing. A microprocessor 41, a ROM storing a control program, which is output to the device;
42, a RAM 43 for data storage, and an I / O circuit 44 for output. Reference numeral 45 denotes a CPU bus through which the microprocessor 41 exchanges data and addresses, control commands and response signals with other parts. In addition, the apparatus of the present embodiment also includes a DC power supply 51, a clock circuit 52 supplied to each unit, and the sensitivity of the imaging unit 10 when the level of the video signal from the imaging unit 10 is low and inappropriate for processing. And a D / A conversion circuit 53 that sends out the signal.

[0005] In order to detect a running vehicle by processing a video signal using such an apparatus, an image type vehicle sensing method generally includes the following steps. That is, a video signal corresponding to an image of an area including a road to be sensed is obtained by an imaging unit such as a CCD camera. In the video signal, a plurality of positions (temporal positions in the video signal) corresponding to measurement points located on a straight line (detection line) in the cross direction of the road are set in advance, and the luminance data groups corresponding to these positions are sequentially set. (See FIG. 6). A luminance data group corresponding to only a road surface on which no vehicle exists on the detection line is extracted at a predetermined time and held as reference road surface level data. The reference road surface level data is corrected as time elapses according to the conditions such as the illuminance of the road (see FIG. 7). The luminance data group along the detection line is compared with the reference road surface level, and if a sufficient difference is recognized (both high and low cases), it is determined that a vehicle is present and a sensing signal is transmitted (see FIG. 7). ).

[0006]

However, when this type of device is actually installed and used on a road, various adverse effects are caused depending on the surrounding environment. One of these is erroneous sensing due to the shadow of the traveling vehicle. That is, when the weather is fine in the daytime, the sunlight is blocked by the vehicle, and a clear shadow (low luminance portion) is formed on the road surface. On the other hand, vehicles to be detected by the apparatus have various paint colors, and there are many vehicles whose luminance is almost the same as that of shadows. Therefore,
As shown in FIG. 6, when the shadow (SD) of the traveling vehicle (CA) is projected on the adjacent lane (R3) and moves, the vehicle sensing device erroneously identifies the shadow of the vehicle as the vehicle and corrects. There was a problem that important information could not be obtained. An effective method of removing the shadow of the vehicle for solving this problem has not been proposed. The present invention has been made in view of the above-described situation, and proposes an image-type vehicle sensing method capable of accurately detecting a vehicle by accurately determining a shadow of a traveling vehicle and processing the shadow as a detection target. That is the task.

[0007]

In order to solve the above-mentioned problems, in an image-type vehicle sensing method according to the present invention, a video signal corresponding to an image of a region including a road is obtained by an image pickup means, and a road signal in the image is obtained. A luminance data group (Dt) consisting of respective luminance levels (Ci) at positions in the video signal temporally corresponding to a plurality of measurement points (Pi) located on a detection line assumed in the transverse direction is sequentially obtained. In response to a case where there is no vehicle on the detection line and only the road surface is present, a luminance data group including a plurality of luminance levels (Ri) is extracted at a predetermined time and held as reference road surface level data (Dr). Dt), and the difference data (Ci-Ri) of each measurement point (Pi) belonging to the predetermined window width Dw section in the road width direction.
Of the variation around the average value La of the differential value (Dfp)
A series of data (differential curve) (Df) corresponding to the measurement line is obtained by sequentially scanning from one end to the other end of the road with the shift amount Dn, and the differential curve (Df) is detected. Threshold level Ts
To obtain a differential binarization curve (Df ′), and smoothing with a spatial smoothing time constant ΔL ′ to obtain spatial smoothing differential binarization data (DfL), and differential data (Ci− Ri)
When a continuous lower value than a certain negative value and the spatial smoothing differential binary signal is “0”, the corresponding portion is determined as a shadow signal (SD ′), and the spatial smoothing differential binary data (DfL) is If there is a sufficient difference from the reference road surface level data (Dr) except for the portion corresponding to the shadow signal (SD '), it is determined that a vehicle is present.

[0008]

According to the above-described method, there is a component that has a luminance level higher than the reference road surface level even in a black vehicle even in fine weather, but has a characteristic that the luminance level is lower than the reference road surface level and hardly changes in the case of a shadow. To use
By determining the shadow of the vehicle and removing it, accurate image-based vehicle sensing processing can be performed.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below in detail with reference to the accompanying drawings. In the method of the present invention, a vehicle as shown in FIG. As shown in FIG. 6, when a signal indicating that a vehicle exists over two lanes is obtained by a shadow during daytime fine weather, whether or not all the obtained signals are actually due to the passage of the vehicle or by a shadow You have to judge what it is. Among vehicles of various paint colors, a vehicle of a white system can be distinguished from a shadow from a difference in luminance level. However, the vehicle of the black system has a low luminance level and a value close to that of the shadow, and it is difficult to distinguish the vehicle from the shadow from the difference in the luminance level. Therefore, the shadow determination in fine weather is to determine whether the vehicle is a black vehicle or a shadow. According to the present invention, as shown in the luminance level distribution diagram of FIG. 7, even in the case of fine weather, even a black vehicle includes components (P1, P2) whose luminance level is higher than the reference road surface level, In this case, the characteristic that the luminance level is lower than the reference road surface level and has almost no change at a substantially constant level is used, and this characteristic is used to distinguish the vehicle from the shadow.

FIGS. 1 and 2 are flow charts showing one embodiment of the method of the present invention. In particular, the algorithm shown in FIG. 1 is characterized by an algorithm for judging the shadow of a vehicle and eliminating erroneous detection due to the shadow. Things. The image-type vehicle sensing method according to the present invention includes the following steps. An image signal corresponding to an image of a region including a road to be sensed is obtained by an imaging unit such as a CCD camera. In the image, a plurality of positions (temporal positions in the video signal) in the video signal corresponding to a plurality of measurement points (Pi) located on a straight line (detection line) assumed in the cross direction of the road are set in advance. , A luminance data group (Dt) composed of luminance levels (Ci) corresponding to these positions is sequentially obtained (S1 · S2,
See FIG. 6). A luminance data group corresponding to only a road surface on which no vehicle exists on the detection line is extracted at a predetermined time and held as reference road surface level data (Dr). The reference road surface level data (Dr) is appropriately corrected with the passage of time according to the conditions such as the illuminance of the road (S6).

Hereinafter, the shadow determination process and the like (S3) will be described as (S31).
To S39). The luminance data group (Dt) along the detection line is processed, and the variation around the average value La of the difference data (Ci-Ri) of each measurement point (Pi) belonging to the window width Dw section in the road width direction is calculated. Obtain as differential value. This scanning is sequentially performed from one end to the other end of the road with a shift amount Dn as shown in FIG. 3, and a series of data groups (differential curves,
Df) is obtained (S31 to S34). The average value La of the difference data (Ci-Ri) in the Dn section is expressed by equation (1) in FIG. 1, and the differential value (Dfp) that is the accumulation of the variation centered on the average value in the Dn section is: It can be obtained by equation (2) in FIG. The differential curve (Df) is binarized on the basis of the detection threshold level Ts to obtain a differential binarized curve (Df ') (S3).
5). Smoothing is performed with the spatial smoothing time constant ΔL ′ to obtain spatial smoothed differential binary data (DfL) (S36). This is a process for removing a differential value corresponding to a boundary (edge) from a shadow to a road surface and the like, and data may protrude. When the difference data (Ci-Ri) is lower than a continuous negative value and the spatial smoothed differential binary signal is "0", the corresponding portion is determined to be a shadow signal (SD ') (S37 to S3).
8). Through each of these steps, the shadow can be extracted by the vehicle.

Spatial smoothed differential binarized data (DfL)
Is compared with the reference road surface level data (Dr) except for the portion corresponding to the shadow signal (SD '), and if a sufficient difference is recognized (both high and low), it is determined that a vehicle is present and the sensing signal is determined. Generate and send (S39, S5). The present invention includes the above-described steps. As described above, it is possible to determine a shadow component generated on a measurement line by a shadow in fine weather as a shadow, remove the shadow component, eliminate erroneous detection, and output only a true vehicle detection signal. The process "development process, (S4)" in FIG. 2 is a process that is performed as necessary. The vehicle is also detected on another measurement line, and the difference between the detection time on both detection lines and the two detection lines are detected. Refers to image processing such as determining the traveling speed of a sensed vehicle from the distance of a line, or extracting and processing an image of a license plate portion after vehicle sensing to identify a vehicle number. Further, extraction of the reference road surface level data can be appropriately selected from various methods, and the extraction or update process is not limited to the position of the embodiment.

[0013]

As described above in detail, according to the image-type vehicle sensing method of the present invention, a video signal corresponding to an image of a region including a road is obtained by an image pickup means, and detection is performed in a direction traversing the road in the image. Multiple measurement points (Pi) located on the line
, A luminance data group (Dt) consisting of respective luminance levels (Ci) at positions in the video signal corresponding to the time is sequentially obtained, the luminance data group (Dt) is processed, and a predetermined window is set in the road width direction. Each measurement point (P
The deviation Dn of the difference data (Ci−Ri) of (i) with respect to the average value La as a differential value (Dfp)
Sequentially scans from one end to the other end of the road to obtain each measurement point, obtain a series of data (differential curve) (Df) corresponding to the measurement line, and detect the differential curve (Df) at the detection threshold level Ts. And differentiated 2
A binarization curve (Df ') is obtained, and a spatial smoothing time constant ΔL' is used to obtain a spatial smoothing differential binarized data (DfL).
When the difference data (Ci-Ri) is continuously lower than a certain negative value and the spatial smoothed differential binary signal is "0", the corresponding portion is determined to be a shadow signal (SD '). Since the vehicle detection is performed except for the portion corresponding to the shadow signal (SD '), the shadow of the vehicle traveling in the adjacent lane is not erroneously detected as the vehicle. ing.

[Brief description of the drawings]

FIG. 1 is a flowchart showing one embodiment of a shadow determination process included in the image-type vehicle sensing method of the present invention.

FIG. 2 is a flowchart showing one embodiment of the image-type vehicle sensing method of the present invention.

FIG. 3 is an explanatory diagram of a shadow determination process included in the image type vehicle sensing method of the present invention.

FIG. 4 is an explanatory diagram of a shadow determination process included in the image type vehicle sensing method of the present invention.

FIG. 5 is a block diagram showing an example of an image-type vehicle sensing device according to the present invention.

FIG. 6 is a diagram illustrating an image related to the image-type vehicle sensing device.

FIG. 7 is a view for explaining a vehicle sensing process and the like of the image-based vehicle sensing method.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 ... Image pick-up means, 20 ... Luminance data conversion part, 30 ... Operation processing part, 40 ... Control part.

Claims (1)

(57) [Claims] An image signal corresponding to an image of a region including a road is obtained by an imaging unit, and a plurality of measurement points (Pi) located on a detection line assumed in a direction transverse to the road in the image are temporally detected. , A luminance data group (Dt) composed of the respective luminance levels (Ci) at the positions in the video signal corresponding to the above is sequentially obtained, and a plurality of luminance levels corresponding to the case where there is no vehicle on the detection line and only the road surface is present. A luminance data group consisting of (Ri) is extracted at a predetermined time and held as reference road surface level data (Dr). The luminance data group (Dt) is processed to measure each measurement point belonging to the predetermined window width Dw section in the road width direction. The variation centering on the average value La of the difference data (Ci-Ri) of (Pi) is used as a differential value (Dfp) as a differential value (Dfp). To obtain a series of data (differential curve) (Df) corresponding to the measurement line, and binarize the differential curve (Df) at the detection threshold level Ts to obtain a differential binarized curve (Df ') To obtain spatial smoothed differential binarized data (DfL) by the spatial smoothing time constant ΔL ′, and the differential data (Ci−Ri) is lower than a continuous negative value, and the spatial smoothed differential When the binary signal is “0”, the corresponding part is determined as a shadow signal (SD ′), and the spatially smoothed differential binary data (DfL) is converted to a shadow signal (Sf).
D ') The reference road surface level data (Dr) except for the equivalent part
An image-based vehicle sensing method, wherein it is determined that a vehicle is present when a sufficient difference is recognized as compared with.