JP2007263714A - Discrimination method and discrimination system of road-surface conditions in winter - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To discriminate road surface conditions with highly accurately, and to constitute inexpensively and operate economically a system. <P>SOLUTION: This discrimination method of the road surface conditions in winter includes a process for determining the concentration mean value by analyzing each distribution of the concentration value and the number of ranges in each fine range on a road surface image in a prescribed range in an image photographed by a television camera 2, and a process for discriminating the road surface condition by determining to which of concentration specified value ranges determined beforehand relative to each road surface condition, the concentration mean value belongs. The method has a characteristic wherein each process is executed by a computer 1. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a winter road surface state determination method and a determination system that automatically determine the state of an entire road surface by evaluating a winter road surface state.

The road surface in the winter in a cold region becomes dry, wet, frozen, snowy, sherbet, etc. There are the following techniques for discriminating such road surface conditions.
(1) The road surface condition is monitored with a temperature sensor embedded in the road surface (see Non-Patent Document 1).
(2) The road surface temperature is detected by the expansion and contraction of an optical fiber embedded in the road surface, and a road freezing danger point is notified (see Non-Patent Document 2).
(3) The road surface height and the amount of reflected light are measured with a light wave sensor, and the road surface temperature is also measured with a temperature sensor to determine the situation such as road surface freezing.
(4) Using a deflection filter, the reflected light from the road surface is separated into vertical deflection and horizontal deflection, and the wet state is detected from the intensity ratio.
http://www.tokimec.co.jp/const/products/eds.htm http://www.hitachi-cable.co.jp/sensor/romen.stm

  However, in all of the above conventional techniques, since the point observation is performed, the amount of information is small, the detection accuracy is low, the reliability is low, and the system operates only in winter. There was a problem that it was uneconomical.

  An object of the present invention is to advantageously solve the above-described problem, and a method for determining a road surface condition in winter according to the present invention is provided for each minute range of a road surface image in a predetermined range in an image photographed by a television camera. Analyzing the distribution of the density value and the number of ranges to determine the average density value, and determining whether the average density value belongs to a predetermined density value range determined for each road condition in advance. And each process is performed by a computer.

  In addition, the winter road surface state determination system according to the present invention is a concentration average that obtains a concentration average value by analyzing the distribution of density values and the number of ranges in each minute range of a road image of a predetermined range in an image photographed by a television camera. And a road surface condition determining means for determining a road surface state by determining to which one of the concentration prescribed value ranges determined in advance for each road surface state the value average means belongs. It is.

  According to the method and system of the present invention, the density average value is obtained by analyzing the distribution of the density value and the number of ranges (for example, the number of pixels) for each minute range (for example, the pixels) of the road surface image in the predetermined range, Since the road surface state is determined by determining which value belongs to the predetermined concentration range determined for each road surface state in advance, the road surface state can be determined with high accuracy.

  Moreover, according to the method and system of the present invention, the road surface state is determined based on the image captured by the television camera, so that the system can be used by utilizing the image of an existing television camera for traffic information used throughout the year. It can be configured at low cost and can be operated economically.

  In the discrimination method and discrimination system according to the present invention, the density average value is any of the four components of the R (red component) image, G (green component) image, B (blue component) image, and luminance image of the road surface image. However, one or more of these components may be obtained and averaged, or the road surface condition obtained from the road surface image of any one of the components may be determined more accurately. A component that can be selected is selected, and a road surface state may be confirmed from a road surface image of the component.

  Further, in the determination method of the present invention, a step of measuring the road surface temperature, and a step of determining the road surface state by determining which of the temperature ranges determined in advance for each road surface state the measured road surface temperature belongs to In the discrimination system of the present invention, the road surface temperature measuring means for measuring the road surface temperature is provided, and the road surface state discriminating means further has the measured road surface temperature determined in advance for each road surface state. The road surface condition may be determined by determining which temperature range it belongs to. In this way, it is possible to more accurately determine freezing and wetting that are difficult to understand only by the concentration.

  Furthermore, in the determination method of the present invention, a step of obtaining an entropy value from a density value and a probability distribution for each minute range of a road surface image of a predetermined range in the image photographed by the television camera, and the entropy value is previously determined on the road surface And determining the road surface state by determining which one of the entropy value ranges determined for each state belongs. In the determination system according to the present invention, a predetermined image in the image captured by the television camera may be included. An entropy value calculating means for obtaining an entropy value from the density value and probability distribution for each minute range of the road surface image of the range, and the road surface state determining means further includes an entropy value range in which the entropy value is predetermined for each road surface state The road surface condition may be determined by determining which one of the two. In this way, it is possible to more accurately determine the snow cover and the sherbet state, which is difficult to understand only by the concentration.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1A is a block diagram showing an embodiment of a winter road surface state discrimination system according to an embodiment of the present invention, which is used in an embodiment of a winter road surface state determination method according to the present invention. ) Is a block diagram showing functions executed by the personal computer of the discrimination system of the embodiment, FIG. 2 is a block diagram showing an application example of the discrimination system of the embodiment, and FIG. 3 is a discrimination system of the embodiment. It is explanatory drawing which illustrates the area | region which discriminate | determines a road surface state.

  As shown in FIG. 1 (a), the winter road surface condition determination system of this embodiment is specifically constituted by a PC (personal computer) 1 that operates according to a predetermined program given in advance. A video (image) signal output from a CCTV (closed circuit television) camera 2 as a television camera that is installed at a predetermined position and photographs a road surface within a predetermined range is distributed and input via a video distributor 3. The video distributor 3 distributes the video signal output from the CCTV camera 2 to the monitor monitor 4. The monitor monitor 4 inputs the video signal output from the CCTV camera 2 and the road surface taken by the CCTV camera 2. Is displayed on the screen.

  The PC 1 also inputs a road surface temperature signal output by a normal road surface temperature sensor 5 as a road surface temperature measuring means embedded in the road surface photographed by the CCTV camera 2 by measuring the temperature of the road surface. Then, the PC 1 functions as a concentration average value calculation unit 1a and a road surface state determination unit 1b by the predetermined program as shown in FIG. 1 (b), and converts the video signal and the road surface temperature signal into field survey data. As described later, the road surface information in the predetermined range in winter is automatically determined as described later, a road surface information signal indicating the determination result is output to the road surface information display device 6 and the road surface information signal is input. The display device 6 outputs road surface information indicated by the road surface information signal on a screen or the like.

  As shown in FIG. 2, the discrimination system of this embodiment is installed by using an existing traffic information management system for displaying traffic images taken by the CCTV camera 2 on the management monitor 4 of the information management room and monitoring traffic conditions. Here, as the road surface information display device 6, in addition to displaying and outputting road surface information on the road information board 6a and the signal light 6b, by pre-recorded text e-mail or pre-recorded voice or synthesized voice automatic call, Road surface information is output by a mobile phone 6c, a personal computer 6d as an external terminal device, a terminal device such as a snow removal base, a car navigation device (not shown) or the like via a modem and the Internet.

By the way, specifically, the discrimination system of this embodiment executes the judgment process according to the following procedure on the basis of a color digital image that is an image taken by the CCTV camera 2.
(1) Setting of measurement area When the PC 1 constituting the discrimination system of this embodiment discriminates the road surface condition, first, as shown by a frame in FIG. 3, the relevant image is displayed in the above image displayed on the screen of the PC 1 A user of the system designates a part of the road surface to be measured as a road surface state measurement area A, and the PCI performs image processing in the measurement area A. The measurement area A can be specified in an arbitrary shape by a normal drawing process.

(2) Subdivision by mesh The set measurement area A is subdivided into a mesh (mesh), and the road surface state is determined for each subdivided area using a specified parameter. In the discrimination system of this embodiment, there are three types of discrimination methods: discrimination by density average value, discrimination by entropy, and discrimination by road surface temperature data. The discrimination system of this embodiment holds the discrimination result for each subdivided area and uses it when discriminating the entire measurement area A.

(3) Image Correction The discrimination system of this embodiment performs two corrections, namely, luminance value correction and image averaging for discrimination of a road surface state to be described later. In the brightness value correction, a predetermined average brightness level of the standard image is acquired, and the brightness level of the discrimination target image is corrected according to the level value. Thereby, the difference in the brightness value at night or in bad weather can be compensated. In addition, since people, cars, etc. are shown in the real-time image of the road, there is a risk that the processing will be hindered. Therefore, in the averaging of images, images for a predetermined period are averaged to create a discrimination target image in which people, cars, etc. are deleted from the image.

(4) Discrimination by Density Analysis of R (Red Component) Image, G (Green Component) Image, B (Blue Component) Image, and Luminance Image The discrimination system of this embodiment is a digital image (in the measurement area A) As shown in FIG. 4, the original image) is separated into four components, an R (red component) image, a G (green component) image, a B (blue component) image, and a luminance image. In addition, when the density level of each component image is divided into five gradations from 0 to 180, for example, the relationship between the density level for each pixel as a minute range and the number of images is examined, and from the result, the density average value (density) Measured values) are calculated respectively. Therefore, PC1 corresponds to a density average value calculation means.

  On the other hand, the discrimination system of this embodiment divides the road surface condition (property) into five types: dry, wet, snow cover, freezing and sherbet, and the above five types for each of the R image, G image, B image and luminance image. For example, the Rth, G, B, and luminance images have the same density average value range (specified density value) for each of the road surface states in the fifth floor from 0 to 180. For each tone level, freeze: 20 to 45 gradations, wet: 50 to 70 gradations, dry: 75 to 95 gradations, sherbet: 100 to 135 gradations, snow cover: 140 to 180 gradations, etc. Pre-set and given.

  Note that the setting of the prescribed density value is performed by comparing the density average value of the photographed image with the actual road surface condition. That is, the inventor of the present application measured the road surface properties at two locations on a certain general national highway for about one month in winter as a verification experiment. During the measurement, the road surface was automatically recorded by the CCTV 2 and the road surface temperature was also measured by the road surface temperature sensor 5. The road surface condition was separately observed by a person so that the validity of the discrimination processing result by the discrimination system of this embodiment could be confirmed. At that time, 150 images of 5 types of images of dry, wet, snow, frozen, and sherbet were extracted from the recorded video, and analysis was performed based on the data of a total of 750 images. FIG. 5 shows the result of determining the road surface state based on the 750 luminance image data, and it was generally possible to determine between dryness, snow accumulation and sherbet from the density average value. On the other hand, the distinction between freezing and wetting, especially during drying, wetting and freezing, was ambiguous. This tendency was the same for the R image, the G image, and the B image.

  Therefore, in the discrimination system of this embodiment, for each area subdivided in the measurement area A, first, the average density value (density measurement value) calculated for the luminance image is the five types of road surface states. Which of the average density value range (specified density value) is checked, and the road surface condition is determined from the result. As a result of the determination, if it is determined that the surface is wet, it is further confirmed that the average density value (concentration measurement value) calculated for the B image enters the wet state among the above five road surface conditions. By doing in this way, the wet state discrimination accuracy can be increased. Accordingly, the PC 1 corresponds to a road surface state determining unit.

(5) Discrimination by entropy In the above-described analysis of the concentration average value, there is a case where the snow concentration and the sherbet have large concentration average values, and the discrimination between the two is ambiguous. Therefore, the discrimination system of this embodiment further converts the color digital image (original image) in the measurement area A into a gray image, and performs the following entropy analysis for each of the subdivided areas. That is, the discrimination system of this embodiment calculates an entropy measurement value in the measurement area A. Therefore, PC1 also corresponds to entropy value calculation means.

Here, entropy refers to an expected value of information amount obtained when an information source is observed, and is also referred to as an average information amount. The entropy is calculated by converting the original color image into a black and white grayscale image (for example, 0 to 255 gradations) as shown in FIGS. 6A and 6B, and calculating from the black and white grayscale image according to the following [Equation 1]. To do.

Where H (X): entropy (bit / pixel), P (xi): probability distribution, and Xi: image density value. If the entropy is large, the data is uniform and corresponds to the sherbet state. Conversely, when the entropy is small, the data is biased and corresponds to a snowy state.

  As shown in FIG. 7, the discrimination system of this embodiment, for example, freeze: 1.5 to 2.4, wet: not to wrap the entropy range (prescribed entropy value) for snow and sherbet. 2.5 to 2.8, dryness: 2.9 to 3.3, snow cover: 3.4 to 4.2, sherbet: 4.3 to 6.0, and the like are set in advance. Then, it is checked whether the entropy measurement value falls within the specified entropy value for snow cover or sherbet, and for determining whether the road surface condition is snow cover or sherbet, the entropy measurement value is given priority over the analysis of the concentration average value. This is performed based on the result of examining which of the above entropy prescribed values falls.

  The setting of the specified entropy value is performed by comparing the measured entropy value of the captured image with the actual road surface condition. That is, FIG. 7 shows the road surface discriminated by calculating entropy based on 750 black and white grayscale images obtained from the 750 color original images. The entropy specified values are: Freezing: 1.5 to 2.4, Wetting: 2.5 to 2.8, Drying: 2.9 to 3.3, Snow cover: 3.4 to 4.2, Sherbet: 4.3 ~ 6.0. In particular, with respect to the discrimination between sherbet and snow cover, as shown in Table 1 described later, the road surface discrimination result based on the above-described luminance image showed that the snow cover was 91% and the sherbet was 93%, while the entropy The result was a higher predictive value, such as 94% snow cover and 97% sherbet.

(6) Improvement of discrimination accuracy by adding road surface temperature In the above-mentioned analysis of the concentration average value, both the freezing and the wetness have small concentration measurement values, which show a similar distribution and are difficult to discriminate. Therefore, the discrimination system of this embodiment discriminates freezing and wetting based on the road surface temperature signal output by the road surface temperature sensor 5 measuring the road surface temperature. That is, here, in addition to the analysis result of the concentration average value, for example, PC1 is frozen when the road surface temperature is negative (below freezing point), and wet when the road surface temperature is 0 ° C. and positive.

  FIG. 8 shows the relationship between the measured density value based on the luminance image and the road surface temperature, and the relationship is further classified by road surface state. As shown in Table 1 to be described later, with regard to freezing and wetting, the road surface discrimination result of only the luminance image showed that the freezing was 75% and the wetting was 85%, while the road surface temperature was taken into consideration (minus temperature is freezing) When 0 ° C. and the plus temperature are set to be wet), it has been found that an improvement in discrimination accuracy can be expected, such as 86% freezing and 85% wetness.

Table 1 below summarizes the road surface discrimination results from the above verification experiment. A discrimination method using a luminance image is effective for drying and a B (blue) image is effective for wetness. Snow cover and sherbet are effective in determining by entropy. As for freezing, adding road surface temperature improves the discrimination system. As described above, it was confirmed in this verification experiment that the road surface can be discriminated with a hit rate close to 90% when data effective for each road surface property is used.
Here, the upper column represents the hit rate at each comparison level, and the lower column represents the number of hits.

  FIG. 9A is an explanatory diagram illustrating the initial state of the discrimination system according to this embodiment. FIG. 9B is an explanatory diagram illustrating the execution state of the discrimination system according to this embodiment. When the determination (discrimination) area is set on the road surface as shown in FIG. 9 (b), the discrimination system of this embodiment subdivides the determination area into an eye shape, and each subdivided area as shown in FIG. The road surface state determined every time is displayed, for example, on a screen with different colors.

  Therefore, according to the winter road surface state determination method of this embodiment and the winter road surface state determination system of this embodiment that implements the determination method, for each area subdivided in the measurement area A By analyzing the distribution of the density value and the number of pixels for each pixel of the road surface image to obtain a density average value, it is determined whether the density average value belongs to a predetermined density value range determined for each road surface state in advance. Since the state is determined, the road surface state can be determined with high accuracy.

  Moreover, according to the discrimination method and discrimination system of this embodiment, the road surface state is discriminated based on the image photographed by the CCTV camera 2, so that an image of an existing TV camera for traffic information used throughout the year can be used. Thus, the system can be configured at low cost and can be operated economically.

  Further, according to the determination method and the determination system of this embodiment, the density average value is the four components of the R (red component) image, G (green component) image, B (blue component) image, and luminance image of the road surface image. First of all, the road surface state is determined using the average density value of the luminance image, and as a result, if there is a component that can be determined more appropriately, the road surface state is confirmed using the concentration average value of that component. It is possible to determine the road surface condition.

  In the discrimination method of this embodiment, the step of measuring the road surface temperature in step (6) above and whether the measured road surface temperature belongs to a temperature range determined in advance for each road surface state is determined. In the discrimination system of this embodiment, the road surface temperature sensor 5 is provided as road surface temperature measuring means for measuring the road surface temperature, and the PC 1 further includes the step (6). Therefore, the road surface state is determined by determining whether the measured road surface temperature belongs to a predetermined temperature range for each road surface state, so that it is possible to more accurately determine freezing and wetting that are difficult to understand only by concentration. Can do.

  Furthermore, in the discrimination method of this embodiment, the density value and probability distribution for each pixel in each area subdivided in the measurement area A in the image taken by the CCTV camera 1 in step (5). And a step of determining the road surface state by determining which entropy value belongs to a predetermined entropy value range determined for each road surface state. In the discrimination system, the PC 1 obtains an entropy value from the density value and the probability distribution for each pixel for each subdivided area in the measurement area A in the image taken by the CCTV camera 1, and further, the entropy is obtained. Since the road surface state is determined by determining which value belongs to the entropy value range determined in advance for each road surface state, Obscure determination of snow and sorbets state can be performed more accurately.

  Although the present invention has been described based on the illustrated example, the present invention is not limited to the above-described example. For example, in the above example, the discrimination area is set only at one place in one image. The discrimination areas may be set at a plurality of locations.

  In the above example, the density average value is obtained for the four components of the R (red component) image, the G (green component) image, the B (blue component) image, and the luminance image of the road surface image. The road surface state may be determined from the concentration average value, or at least one of the determination based on the entropy value and the determination based on the road surface temperature measurement value may be combined.

  Thus, according to the method and system of the present invention, the density average value is obtained by analyzing the density value and the distribution of the number of ranges (for example, the number of pixels) for each minute range (for example, the number of pixels) of the road surface image in the predetermined range. Since the road surface state is determined by determining which value belongs to the predetermined concentration range determined for each road surface state in advance, the road surface state can be determined with high accuracy.

  Moreover, according to the method and system of the present invention, the road surface state is determined based on the image captured by the television camera, so that the system can be used by utilizing the image of an existing television camera for traffic information used throughout the year. It can be configured at low cost and can be operated economically.

(A) is the block diagram which shows one Example of the discrimination | determination system of the road condition of the winter of this invention used for one Example of the discrimination method of the road condition of the winter of this invention, (b) is the structure of the Example It is a block diagram which shows the function which the personal computer of a discrimination | determination system performs. It is a block diagram which shows the application example of the discrimination | determination system of the said Example. It is explanatory drawing which illustrates the area | region where the discrimination | determination system of the said Example discriminate | determines a road surface state. It is explanatory drawing which shows the density distribution example of each component of the color image which the discrimination | determination system of the said Example analyzes. It is explanatory drawing which shows the discrimination | determination result of the road surface state by the luminance image which the discrimination | determination system of the said Example performed. It is explanatory drawing which shows the example of a monochrome grayscale image for the entropy analysis which the discrimination | determination system of the said Example performs. It is explanatory drawing which shows the discrimination | determination result of the road surface state by the entropy analysis which the discrimination | determination system of the said Example performed. It is explanatory drawing which shows the relationship between the density | concentration measured value by the luminance image which the discrimination | determination system of the said Example calculated | required, and road surface temperature. (A) is explanatory drawing which illustrates the initial state of the discrimination | determination system of the said Example, (b) is explanatory drawing which illustrates the execution state of the discrimination | determination system of the Example.

Explanation of symbols

1 Personal computer (PC)
DESCRIPTION OF SYMBOLS 1a Concentration average value calculating part 1b Road surface state discrimination | determination part 2 CCTV camera 3 Video distributor 4 Monitoring monitor 5 Road surface temperature sensor 6 Road surface information display apparatus 6a Road information board 6b Signal light 6c Mobile phone

Claims (6)

Analyzing the density value for each minute range and the distribution of the number of ranges of the road surface image of the predetermined range in the image taken by the TV camera to obtain a density average value;
Determining the road surface state by determining which concentration average value belongs to which of the concentration prescribed value ranges determined in advance for each road surface state,
A method for determining a road surface condition in winter, wherein a computer performs each of the steps. Measuring the road surface temperature;
A step of determining the road surface state by determining which of the temperature ranges that the measured road surface temperature belongs in advance for each road surface state;
The method for determining a road surface condition in winter according to claim 1, wherein: A step of obtaining an entropy value from a density value and a probability distribution for each minute range of a road surface image of a predetermined range in an image photographed by the television camera;
The winter road surface according to claim 1, further comprising a step of determining a road surface state by determining which entropy value range belongs to a predetermined entropy value range for each road surface state. How to determine the state. A density average value calculating means for analyzing a density value for each minute range and a distribution of the number of ranges of a road image in a predetermined range in an image photographed by a television camera to obtain a density average value;
Road surface state determining means for determining the road surface state by determining which of the concentration specified value ranges the concentration average value is determined in advance for each road surface state;
A road surface condition discrimination system in winter, characterized by comprising: Provide road surface temperature measuring means to measure the road surface temperature,
5. The winter period according to claim 4, wherein the road surface state determining means further determines the road surface state by determining which of the temperature ranges determined in advance for each road surface state the measured road surface temperature belongs to. Road surface condition discrimination system. An entropy value calculating means for obtaining an entropy value from a density value and a probability distribution for each minute range of a road surface image of a predetermined range in an image photographed by the television camera;
6. The road surface state determining means further determines the road surface state by determining which entropy value belongs to a predetermined entropy value range determined for each road surface state. A winter road surface classification system.