JPH02230499A - Traffic flow measuring device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a traffic flow measuring device for road vehicles, and in particular, it predicts qualitative characteristics such as congestion and provides accurate information to road managers and users. The present invention relates to a traffic flow measuring device suitable for providing traffic information.

[Background Art] With the development of automobile transportation, the need for traffic congestion prediction on expressways, travel time prediction services, and route guidance is increasing. In order to respond to such situations, a traffic measurement system has been developed that uses not only vehicle sensors but also television cameras that can be visually confirmed to measure traffic conditions as images and provide users with more accurate information. (Japanese Unexamined Patent Publication No. 61-90330, etc.).

This type of system captures an overhead image of the road, extracts and recognizes only vehicles from this image, and
It measures the number of vehicles, individual vehicle speeds, vehicle types, and occupancy rates, and uses this data to obtain a comprehensive understanding of traffic flow. In such systems, various methods have been developed for vehicle extraction and recognition from input images, but they only go beyond measurement and have little ability to predict future traffic changes from the current state. No efforts were made to utilize the information and provide useful forecast information online to road administrators and users (Japanese Patent Laid-Open No. 6
3-35281 etc.).

In particular, the number of long tunnels is rapidly increasing due to the extension of roads into mountainous areas and the need for countermeasures against overcrowding and noise in urban areas, but these long tunnels are mainly monitored visually using industrial television cameras. This puts a heavy burden on supervisors who are forced to be under stress for long periods of time.

In addition, conventional methods for detecting congestion in traffic flow include methods in which vehicle speeds and occupancy rates detected by loop-type or ultrasonic sensor-type sensors are differentiated using threshold values, and the congestion level is determined based on the results or a combination of results. There is a method for determining and predicting congestion by focusing on compression waves, which is one of the phenomena peculiar to congested traffic flow (Japanese Patent Application Laid-Open No. 62-295200, Special 1m Sho 63-5).
3698 etc.).

These methods process measured values using linear equations, etc., and detect and predict traffic jams using positive and negative fluctuation factors found by comparing them with a certain threshold. They were not necessarily well-suited to traffic jams.

[Problems to be Solved by the Invention] In the above-mentioned conventional technology, the threshold value and compression waves, which are one of the congestion phenomena, are mathematically processed based on the measured value of the sensor, and congestion etc. are predicted. However, there was a problem in that there was no consideration given to qualitatively understanding and analyzing the various factors that caused traffic jams, making it easy for errors to occur.

The purpose of the present invention is to treat various detected values of spatial state quantities (traffic volume, vehicle speed, occupancy rate, large vehicle mix rate, inter-vehicle distance, etc.) through image processing as ambiguous quantities, and to It is an object of the present invention to provide a traffic flow measuring device capable of detecting and predicting traffic jams, displaying information, and outputting warnings.

[Means for Solving the Problems] In order to achieve the above object, the present invention performs image processing on traffic flow image data from an image input device to extract and recognize vehicle data, and calculates the number, speed, and occupancy rate of vehicles. In the traffic flow measurement device that measures traffic flow information such as congestion and outputs traffic flow information such as traffic jams, a plurality of auto7I+remodules each measure the number, speed, occupancy rate, etc. of vehicles extracted and recognized through image processing,
il81! A differentiation module that differentiates the l result from the first to the nth order, a qualitative fuzzy inference module that takes in the I measurement result and the differentiation result, executes at least one stage of fuzzy inference, and outputs a fuzzy evaluation set for the control amount, and a fuzzy evaluation The present invention proposes a traffic flow measurement device that includes a calculation module that takes in the set and derives a quantitative value of the control amount.

Even if a plurality of image input devices are arranged at remote locations, the overall configuration can be simplified by installing the image processing device and other parts in a central control facility as common processing means for the plurality of image input devices. .

[Operations] In the present invention, the measured value from the image is taken as an ambiguous quantity, the traffic flow is grasped as a qualitative movement on an n-dimensional road map, and the overall traffic flow is calculated by fuzzy inference. Evaluate trends and perform accurate congestion detection and prediction.

Therefore, it can better correspond to changes in actual traffic flow,
Since the predictions are accurate, there is almost no inconvenience like in the past, where you go to a spot where there is a traffic jam sign only to find out that there is no traffic jam at all.

In particular, it is now possible to output predictive information and warnings online when abnormalities such as traffic jams are predicted, without having to continue visual monitoring using TV cameras for monitoring long tunnels, etc., reducing the burden of monitoring work. It can be significantly reduced.

[Example] Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of a traffic flow measuring device according to the present invention. The traffic flow measurement device 1 captures images from the image input device 2 at, for example, a fixed period, and measures the traffic flow. The image input device 2 is a television camera or the like installed around the road, and captures images of vehicles flowing on the road.

The traffic flow measuring device which is the object of the present invention consists of an image processing module 3 , a dividing device 4 , a storage device 9 , a display device 10 , a display 11 , and an output device 12 . The cutting device 4 includes a plurality of measurement modules 5, a differential (difference) module 6, a qualitative fuzzy inference module 7, and an arithmetic module 8.

The image processing module 3 takes in image data from the image input device 2, extracts and recognizes a vehicle.

An example of the internal configuration of the image processing module 3 is shown in FIG. The image processing module 3 first obtains a difference image from the input image and a background image in which no vehicle is shown. Next, a binary image is obtained by distinguishing the difference images using threshold values for gradation. When obtaining a binary image from this difference image, processing such as differentiation, smoothing, enlargement/reduction, etc., may be performed on the difference image. Furthermore, vehicle features are extracted and recognized for each driving lane from the binary image.

The measurement module 5 in the measurement device 4 measures the number of vehicles, vehicle speed, vehicle type, occupancy rate, etc. with each module.

The differentiation (difference) module 6 differentiates the cutting values from the measurement module 5 from the first to the nth orders.

The qualitative fuzzy inference module 7 has an internal configuration as shown in FIG.
One-stage or multi-stage fuzzy inference is performed as an ambiguous quantity, and a fuzzy evaluation set (a-Z) for the control quantity is output. The fuzzy evaluation set is a set of fitness degrees (0.0 to 1.0) for each evaluation of the control amount.

The calculation module 8 takes in the fuzzy 7-evaluation set (a - z) for the controlled variable, cuts off the evaluation function of each controlled variable by the fuzzy set value, as shown in FIG. 4, and further calculates the center of gravity for the effective area. etc.,
Introduce quantitative values of control variables.

The storage device 9 is connected to the image processing module 3 to the calculation module 8, the display device 10, and the output device 12, and stores input images, measured values, differential values, qualitative variables, fuzzy evaluation values (membership functions), etc. do.

The display device 10 extracts necessary data and images from the contents of the storage device 9 and outputs and displays them on the display 11.

The output device @12 displays the contents of the storage device 9 (especially control quantitative values)
This is an interface that outputs the information to an external medium or communicates with an external device.

In the traffic flow measuring device of the present invention having the above configuration, the traffic flow (
By measuring the number of vehicles (number of vehicles), vehicle speed, occupancy rate, vehicle type (rate of dog-shaped vehicles), distance between vehicles, etc., it is possible to detect congestion and accurately predict traffic volume in the near future.

Factors that affect traffic on actual roads include road regulations such as construction and the installation environment of traffic flow measuring devices, but here we will use typical traffic volume, vehicle speed, and occupancy rate as examples. explain.

Figure 5 is a diagram showing the correlation between traffic density (occupancy rate) and speed.
FIG. 6 is a diagram showing the correlation between traffic density (occupancy rate) and traffic volume, and FIG. 7 is a diagram showing the correlation between speed and traffic volume. In these correlation diagrams, the area where vehicles can flow freely is defined as the free flow area, and the area where vehicles are restrained by the vehicle in front resulting in unstable unsteady flow and the area where congestion occurs is defined as the congestion area. .

If we represent the correlations in Figures 5 to 7 on a three-dimensional space, we can express the congestion area as a specific area on an overhead view and grasp the position and direction within the area, as shown in Figure 8. If so, it is possible to accurately determine traffic jams. In addition, although FIG. 8 expresses the above-mentioned three important factors, if there are more factors, it can be expanded to n dimensions.

An example of qualitative fuzzy inference is shown in FIG. Here, for ease of understanding, one factor of traffic density (occupancy rate) is shown.

In the same figure (A), for example, when the occupancy rate is 30%,
Let H be the qualitative evaluation function that determines the occupancy rate (○) as high, M as the qualitative evaluation function that determines it as medium, and L as the qualitative evaluation function that determines it as low. When the occupancy rate is 30%, it is high (○).
) u is 0.6, medium (0) M is 0.8, low (○) 1
, is evaluated as 0.3.

In addition, in the same figure (B), when the rate of change in occupancy rate a○ is evaluated using a qualitative evaluation function of decrease -, unchanged O, and increase 0, when it changes by -5%, the target value (for example, congestion)
The fitness of increasing (a○) 10 is evaluated as 0.2, unchanged (a○) 0 is evaluated as 0.8, and decreasing (aO) is evaluated as 0.5.

Here, each fuzzy inference has the following relationship when, for example, the minimax method is applied, and can be defined by the causal structure shown in FIG. 9(C).

This causal structure defines a qualitative state based on FIG. The rule is 3 states (H, M.
L) and the occupancy differential value (a O) in three states (-, 0, 10)
There are a total of 9 combinations.

For example, λ, occupancy rate (0) is high < (H). If the change (a○) decreases (1), it means that the amount of change in traffic congestion (aρ) decreases.

If you write this in if-then format,
jf ○ is L ka a○ is then aρ becomes.

Specifically, the goodness of fit shown in FIG. 9 is as follows: μH (0) , μs (0) , μL (0) =
(0.6, 0.8, 0.3) μ+ (ao) .
uo (ao). tt-(ao)(0.2,
0.8, 0.5), and the if-then rule can be expressed as follows.

(aρ)=Mj. n(μ(O),μ(0)M]n
(0.3, 0.5) 0.3 Similarly, the fitness of the other 8 rules is calculated using the 10th
As shown in the figure.

A comprehensive fuzzy evaluation set of aρ is determined from the evaluation values shown in FIG. 10 by + (increase), 0 (unchanged), and -(?ffC decrease). In other words, from Figure 10, the set of fitness is 4 rules for increase, 1 rule for unchanged, and 4 rules for decrease, and tt+(aρ) = Max (Min (μo (0
) + μ+ (ao) )M]n (μo (0)
+ po (a O) ) Min (ttM(0) +
μ+ (ao) ) Min(μ+, (0) r μ+
(a O) )=Max (0.2, 0.6, 0
．． 2, 0.2)=0.6 Similarly, μo(aρ)=0.8, μ(aρ)0.5
becomes.

Therefore, the qualitative fuzzy evaluation suitability set S for aρ is S(aρ)=(μ+(aρ), μo(aρ), μ−(a
ρ)) = {0.6, 0.8, 0.5}, and a qualitative fuzzy inference result is obtained.

Similar calculations can be performed not only for occupancy, but also for factors such as degree.

Based on the qualitative fuzzy inference results obtained in this manner, the calculation module 8 executes an inverse transformation calculation using a predetermined qualitative evaluation membership function, and an example of obtaining a control quantitative value (congestion level) is shown in FIG. 11. . The calculation module 8 increases (+
), unchanged (0), and decreased (1), the degree of congestion is obtained by performing a cut-off using each suitability value of the suitability set S, and calculating the center of gravity.

In this way, according to this embodiment, it is possible to qualitatively evaluate movement as a whole based on the occupancy rate, etc., and quantitatively determine and predict the degree of congestion.

The above example has been simplified in order to clarify the operations of the qualitative fuzzy inference module 7 and the calculation module 8 such as traffic congestion change suitability. An example of the overall structure of a mechanism for predicting and inferring actual traffic congestion is shown in FIG. Figure 12 shows
Descriptions are made corresponding to each step and number of steps in FIG.

In this example, as already mentioned, the congestion degree prediction is not estimated based on a single factor, but is evaluated based on a comprehensive judgment based on multiple factors and a comprehensive understanding. In other words, in addition to the amount of change in congestion degree (a congestion degree) due to the correlation between the estimated values of traffic volume and occupancy rate, traffic volume and speed, and speed and occupancy rate,
The amount of change in congestion (a congestion degree) is comprehensively inferred from the amount of change in congestion degree (a congestion degree) due to the correlation between traffic volume, inter-vehicle distance, occupancy rate and its rate of change (a occupancy rate), and the goodness of fit is calculated. Ask qualitatively.

According to this method, it is possible to grasp the traffic volume from various factors in a comprehensive manner, and it is possible to accurately predict the degree of congestion, etc.

[Effects of the invention] According to the present invention, in the image-based traffic flow measuring device,
Taking measured values as ambiguous quantities, analyzing traffic flows as qualitative movements on an n-dimensional elevation, evaluating the overall trend of traffic flows using fuzzy inference, and accurately detecting and predicting traffic jams. Moreover, a traffic flow measurement device that can be quickly executed can be obtained.

In particular, it is now possible to output predictive information and warnings online when abnormalities such as traffic jams are predicted, without having to continue visual monitoring using surveillance cameras such as long-running tunnels, greatly reducing the burden on monitoring operations. can be reduced to

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment of a traffic flow measuring device according to the present invention, FIG. 2 is a block diagram showing an example of an image processing module, and FIG. 3 is a block diagram showing an example of the configuration of a qualitative fuzzy inference module. Block diagram, Figure 4 is a diagram showing the functions of the calculation module, Figure 5 is a diagram showing the correlation between traffic density (occupancy rate) and speed, Figure 6 is a diagram showing the correlation between traffic density (occupancy rate) and traffic volume. Figure 7 is a diagram showing the correlation between speed and traffic volume, Figure 8 is a diagram showing the traffic flow correlation three-dimensionally and in an overview, and Figure 9 is an example of qualitative fuzzy inference. 10 is a diagram showing an example of traffic congestion change adaptability, FIG. 11 is a diagram showing an example of arithmetic module processing, and FIG. 12 is a diagram showing the overall structure of the congestion prediction inference function. ]--...Traffic flow measurement device, 2...Image input device, 3.
Image processing module, 4 Measuring device, 5...Measuring module/Differential (difference) module, Qualitative fuzzy inference module, Arithmetic module, 9...Storage device, ○...Display device, 11.Display device, 2...・Output device. Agent Tatsu Unuma Figure H7”J 口tJfifh re'1 g2OH, (2
M.

Claims (1)

[Claims] 1. Extract and recognize vehicle data by image processing traffic flow image data from an image input device, measure the number of vehicles, speed, occupancy rate, etc., and output traffic flow information such as congestion. In the traffic flow measurement device, the number, speed, and number of vehicles extracted and recognized by the image processing are
a plurality of measurement modules that respectively measure occupancy rates, etc.; a differentiation module that differentiates the measurement results from the first order to the nth order; and at least one differentiation module that takes in the measurement results and the differentiation results.
A qualitative fuzzy inference module that executes fuzzy inference of stages and outputs a fuzzy evaluation set for a controlled variable; and an arithmetic module that takes in the fuzzy evaluation set and derives a quantitative value of the controlled variable. Measuring device. 2. In the traffic flow measurement device according to claim 1, a plurality of the image input devices are arranged remotely, and means below the image processing device are installed in a central control facility as processing means common to the plurality of image input devices. A traffic flow measuring device characterized by being installed.