JP2010128732A - Apparatus for observing density in the number of vehicles and program for the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an apparatus for observing density in the number of vehicles, which can automatically and accurately recognize each vehicle as a vehicle line segment from a satellite image or the like whose resolution is low and accurately observe density in the number of vehicles in each local area. <P>SOLUTION: The apparatus 1 for observing density in the number of vehicles includes: a keyboard 4; a processor 2 comprising an image input part 11, a control part 10, a road area extraction part 12, a local area division part 13, a road direction calculation part 14, a segment group extraction part 15 for extracting a line segment group from the whole input image, a vehicle line segment extraction part 16 for extracting only a vehicle line segment equivalent to the front or rear of a vehicle from the found line segment group, a number-of-vehicles density calculation part 17 for calculating the density of a vehicle segment group in each divided local area, and a number-of-vehicles density display processing part 18 for performing image processing for combining the calculated density in the number of vehicles with corresponding map information; and a display device 5 for displaying an image processing result. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a vehicle number density observation device and a program for the vehicle number density observation device, and more specifically, a novel and novel vehicle number density observation device for calculating a vehicle number density on a road using an image taken from above. The present invention relates to a program for the vehicle number density observation apparatus.

  Understanding the traffic situation on the road in a wide area is important in the investigation and analysis of traffic jams and parked vehicles. In order to grasp this traffic situation, observation devices such as detectors and surveillance cameras installed in the vicinity of the road are generally used, and the results are obtained by integrated calculation of locally collected information.

  However, with such a method, the measurement accuracy differs depending on the installation interval of the observation equipment, it is difficult to simultaneously grasp the number of vehicles in a wide area at the same time, and if the equipment is damaged or stopped due to a large-scale disaster, It becomes impossible to grasp the traffic situation itself.

  On the other hand, methods of using images taken from the sky such as satellite images and aerial images are used in various fields for analysis of the ground surface in a wide area.

  For example, a device that recognizes road information from a satellite image (for example, Japanese Patent Application Laid-Open No. 2007-33931: a road recognition system for creating a map using a satellite image or the like) or a device that detects a disaster situation in a disaster (for example, a special feature) No. 2008-107941 gazette: monitoring device).

  Several studies have already been proposed that use images taken from the sky to grasp traffic conditions.

  For example, one image is used to extract the position of each vehicle (for example, extraction of a vehicle area in an urban area using aerial images, Kumiko Masuyama, Saji Hajime, ViEW2005 Vision Technology Actual Use Workshop Proceedings , Pp.178-181, (2005), vehicle detection using satellite images, Fumiaki Sato, Saiharu Saji, MIRU2008, Proceedings of Symposium on Image Recognition and Understanding, pp.532-536, (2008)), or What calculates | requires the motion of each vehicle using the several image disclosed by patent document 1 (Unexamined-Japanese-Patent No. 2005-56186: It is a traffic condition observation system etc. Furthermore, as related technology, it is an advanced imaging time series. Construction of vehicle motion recognition method using images, Takashi Fuse, Hidenori Shimizu, Ryo Maeda, Journal of Japan Society of Civil Engineers, IV-60, No.737, pp.159-173, 2003) are known.

  However, any of the above-described methods has been proposed on the premise that an image that can clearly recognize the position and shape of each vehicle is used.

  Such a premise can only be realized under limited ideal conditions. For example, from a satellite image with a low resolution or an image in which the shape of the vehicle is unclear due to the influence of the shooting environment (weather, shooting date, shooting angle, etc.) It is difficult to accurately recognize one vehicle at a time in terms of both calculation accuracy and calculation time.

JP 2005-56186 A

  The problems to be solved by the present invention include, for example, one vehicle and one vehicle from low-resolution satellite images and images whose vehicle shape is unclear due to the influence of the shooting environment (weather, shooting date, shooting angle, etc.). There is no new or novel vehicle number density observation apparatus that can automatically recognize the vehicle line segment accurately and accurately observe the vehicle number density for each local region.

  A vehicle number density observation apparatus according to the present invention is a vehicle number density observation apparatus that observes a vehicle number density on a road using an input image based on an image taken from the sky and map information. Means for superimposing the map information; means for extracting only the road area from the superimposed information; means for dividing the extracted road area into local areas; and means for determining the road direction for each of the divided local areas; Means for extracting a line segment group from the entire input image; means for extracting only the vehicle line segment group corresponding to the front or rear surface of the vehicle from the obtained line segment group; and the density of the vehicle line segment group for each divided local region. The most important feature is that it has means for obtaining, means for performing image processing for synthesizing the obtained vehicle number density on the corresponding map information, and presentation means for visually presenting the image processing result. That.

  According to the first to ninth aspects of the present invention, a line segment group corresponding to a vehicle is collectively extracted from a road area in an image photographed from above, and the density of the line segment group in the road area is calculated. Since the vehicle density in the road area is obtained and presented visually, it is not necessary to accurately extract each vehicle, so even if the shape of the vehicle is unclear, it is simple. It is possible to realize and provide a vehicle number density observation apparatus capable of automatically observing the degree of congestion of vehicles on a wide area road with simple calculation.

  According to the tenth aspect of the present invention, it is not necessary to accurately extract each vehicle using a general-purpose computer, and even if the shape of the vehicle is unclear, simple calculations can be performed on a wide area road. It is possible to realize and provide a program for a vehicle number density observation apparatus capable of automatically observing the degree of congestion of vehicles.

  The present invention automatically and accurately recognizes each vehicle as a vehicle line segment from low-resolution satellite images and images whose vehicle shape is unclear due to the influence of the shooting environment (weather, shooting date, shooting angle, etc.). The object of the present invention is to realize and provide a novel and novel vehicle number density observation apparatus capable of accurately observing the vehicle number density for each local region.

  A vehicle number density observation apparatus according to the present invention is a vehicle number density observation apparatus that observes the number density of vehicles on a road using an input image based on an image taken from the sky and map information, and performs overall control. From the superimposed information, an input means for inputting an image file name and the like, an image input unit for inputting an input image and map information, an overlay processing unit for overlaying the input image and the map information, and the superimposed information A road area extracting unit that extracts only a local area of the road, a local area dividing unit that divides the extracted road area into local areas, a road direction calculating unit that obtains a road direction for each divided local area, and an input image A line segment group extraction unit that extracts line segment groups from the whole, a vehicle line segment extraction unit that extracts only a vehicle line segment group corresponding to the front or rear surface of the vehicle from the obtained line segment group, and a vehicle line for each divided local region Min Vehicle number density calculating unit for determining the density of the vehicle, a processing device having a vehicle number density display processing unit for performing image processing for synthesizing the determined vehicle number density on the corresponding map information, and a display device for displaying the image processing result The above object is realized by a configuration having

Hereinafter, a vehicle number density observation apparatus according to an embodiment of the present invention will be described in detail with reference to FIGS. 1 to 10.
As shown in FIG. 1, the vehicle number density observation apparatus 1 according to the present embodiment is configured by using, for example, a general-purpose personal computer, and is an input (upper sky) image (hereinafter referred to as an “upper sky” image). , "Input image") and map information M prepared in advance are used to execute various processes for determining the number of vehicles in a local area on a wide road under the control of the control unit 10 storing an operation program Processing device 2, storage device 3 that stores input image I, map information M, and the like, keyboard 4 that is an input means for inputting image file names and various commands, and vehicle calculated by processing device 2 And a display device 5 including a liquid crystal display for visually presenting the number density and the like.

  The input image I in this embodiment is, for example, a 1 m resolution IKONOS satellite image ((C) Nippon Space Imaging Co., Ltd.), and the map information M used is, for example, a high-precision vector with a scale of 1/2500. Map Tokyo V Map ((C) Mues Co., Ltd.).

  The processing device 2 includes a control unit 10 that performs overall control, an image input unit 11 that inputs an input image I and map information M, and an overlay processing unit 19 that overlays the input image I and map information M. A road area extracting unit 12 that extracts only a local area of the road from the superimposed information, a local area dividing unit 13 that divides the extracted road area into local areas, and a road direction is obtained for each of the divided local areas. A road direction calculation unit 14, a line segment group extraction unit 15 that extracts a line segment group from the entire input image, and a vehicle line segment extraction unit 16 that extracts only a vehicle line segment group corresponding to the front or rear surface of the vehicle from the obtained line segment group. And a vehicle number density calculation unit 17 for obtaining the density of the vehicle line segment group for each divided local region, and a vehicle number density display processing unit 1 for performing image processing for combining the obtained vehicle number density on the corresponding map information. And, it is equipped with.

  The image input unit 11, the road region extracting unit 12, the local region dividing unit 13, the road direction calculating unit 14, the line segment group extracting unit 15, the vehicle line segment extracting unit 16, the vehicle number density calculating unit 17, and the vehicle number density display processing unit. 18 and the overlay processing unit 19 can be configured by a program group for performing each processing as described below.

Next, a processing flow of the vehicle number density observation apparatus 1 according to the present embodiment will be described.
First, the image input unit 11 is used to input the sky image as shown in FIG. 3 as the input image I and also input the map information M shown in FIG. 4 corresponding to the input image I (step S101) and store it. Store in device 3. Note that FIG. 3 is a 1 m resolution IKONOS satellite image ((C) Nippon Space Imaging Co., Ltd.). FIG. 4 shows a high-precision vector map Tokyo V map ((C) Mues Co., Ltd.) with a scale of 1/2500.

  The target sky image here is an image taken with a satellite, Cessna, helicopter, balloon, etc., or an image taken with a camera built in a high-rise building, etc. The image quality is about a few pixels of the vehicle However, it is not necessary to have such a clearness and resolution that the vehicle shape itself can be clearly recognized.

  Next, a process of superimposing (positioning) the input image I and the map information M using the superimposition processing unit 19 is performed (step S102), and the map information M using the road region extraction unit 12 is further performed. A road area image Iload in which only the information in the road area is left is created by distinguishing the road area and the area outside the road in step S103.

  Here, the input image I and the map information M are overlapped by a known method (for example, extracting a plurality of corresponding points from the input image I and the map information M, and solving a geometric transformation expression such as projective transformation). Adopt the method of superposition.

  Then, in the processing by the road area extraction unit 12, the pixels of the input image I located at the position corresponding to the road area of the map information M are regarded as pixels constituting the road area, and other pixels are left as they are. As a result, a road area image Iload in which information is deleted (for example, black) is created.

  Next, the local area dividing unit 13 is used to divide the road area in the road area image Iload into local areas Ri (I = 1, 2, 3,...) (Step S104).

The division here may be performed for each region unit for which it is desired to grasp the traffic situation. As the boundary, for example, as shown in FIG. 5, an intersection, a median strip, a center line, and the like are candidates.
The information on the intersection and the median strip is read from the superimposed map information M.

  Next, the road direction calculation unit 14 is used to calculate the road direction RAi (x, y) for each pixel (x, y) in the local region Ri (step S105).

  The road direction RAi (x, y) here corresponds to the traveling direction of the vehicle. Specifically, as shown in FIG. 6, the road direction RAi (x, y) is from the direction of the boundary line of the local region Ri parallel to the road boundary. It is obtained by interpolating the direction of each pixel in the region. FIG. 6 shows an example in which the local region Ri is rectangular. The same calculation is performed when the local region Ri is curved. For the local area Ri corresponding to the intersection, the road direction is undefined.

  Next, the line segment group extraction unit 15 is used to extract a line segment (edge) group from the input image I (step S106).

  A specific method of line segment group extraction processing in this case will be described in detail with reference to FIG. 7. Edge extraction processing is applied to the entire input image I. For example, a technique such as a well-known Sobel operator (a method of obtaining a gradient intensity (differential value) by local product-sum operation) is used as an image edge extraction method.

  Then, those obtained by connecting the obtained pixels having high edge strength are extracted as line segment candidate areas.

  Furthermore, from the line segment candidate areas, those that do not have a straight line shape, those that are extremely long or short compared to the size of the vehicle, and those that correspond to the boundary of the shadow area are deleted, and the remaining areas are removed. Extract as a line group.

  A known method is used for determining whether or not the shape is a straight line shape. Further, in determining whether or not the boundary corresponds to the boundary of the shadow area, each pixel of the input image I is divided into a shadow area and a sunny area where both the brightness and the saturation are both low. It is determined whether or not the line segment of interest is located on the boundary line of the shadow area.

  Referring to the edge extraction method in the shadow area, in consideration of the fact that the edge intensity is lower in the shadow area than in the sunny area, the edge intensity threshold is set lower in the shadow area and extracted.

  Next, as shown in FIG. 8, the vehicle line segment extraction unit 16 is used to delete the line segment group of the input image I that corresponds to any one of the following a and b, and the remaining ones. A line segment (vehicle line segment) corresponding to the front or rear surface of the vehicle is extracted (step S107).

  That is, a line segment that exists in the local region Ri and is perpendicular to the road direction RAi (x, y) is extracted as a vehicle line segment corresponding to the front or rear surface of the vehicle.

That is,
a. The whole or part of the line segment is outside the local region Ri (Ri = 1, 2, 3...).
b. The road direction RAi (x, y) at the pixel at the center point of the line segment is not perpendicular to the line segment direction.

  Here, a line segment corresponding to “a” corresponds to a building outside the road, an auxiliary facility near the road, and the like, and a line segment corresponding to “b” corresponds to a sign on the road, a side surface of the vehicle, or the like. Note that there is no line segment to be deleted under the condition b in the local area corresponding to the intersection (because the road direction is undefined at the intersection).

Next, as indicated by numerical values in FIG. 9, the vehicle number density calculating unit 17 is used to calculate the vehicle number density RDi from the vehicle line segment in the local region Ri (step S108).
The vehicle line segment calculated in step S108 is located in a pixel corresponding to the front or rear surface of the vehicle, and this number corresponds to the number of vehicles, so the vehicle number density RDi can be obtained from the vehicle line segment. In FIG. 9, the higher the numerical value, the higher the number of vehicles on the road.

  The vehicle number density RDi is obtained by dividing the vehicle line segment NRi in the local region Ri by the area (number of pixels) SRi of the local region Ri, that is, by calculating RDi = NRi / SRi.

  Next, image processing for displaying the vehicle number density RDi obtained using the vehicle number density display processing unit 18 on the map information M is performed (step S109).

In this display process, for example, the vehicle number density RDi is divided into stages by some threshold values, and is displayed by being color-coded for each local area Ri on the map information M, or the local area Ri is collectively wider. It is possible to realize a processing mode according to the application, such as displaying for each road area.
Then, the processing result of the vehicle number density display processing unit 18 is displayed on the screen of the display device 5 as shown in FIG. 10 (step S110).

  FIG. 10 shows an example of color-coded display for each local area Ri on the map information M. In FIG. 10, for convenience of explanation, the darker the color, the larger the vehicle number density RDi, that is, the vehicle It indicates that there is traffic.

  According to the vehicle number density observation apparatus of the present embodiment described above, the shape of the vehicle is unclear due to the influence of the input image I such as a low-resolution satellite image or the shooting environment (weather, shooting date, shooting angle, etc.). Realize and provide a new and innovative vehicle number density observation device that can automatically recognize each vehicle on the road as a vehicle line segment accurately and accurately observe the vehicle number density in each local area from a simple image Can do.

  Further, according to this embodiment, it is not necessary to accurately extract each vehicle using a general-purpose computer, and even if the vehicle shape is unclear, the vehicle on the wide area road can be calculated simply. It is possible to realize and provide a new and novel program for the vehicle number density observation apparatus 1 that can automatically observe the degree of congestion.

  In addition to the case described above, the present invention inputs a specific road section for calculating the corresponding point of the sky image and the map information and the vehicle density, and is configured to automatically calculate the vehicle density in the section. Therefore, it is possible to collect vehicle density information in a wide area and to collect traffic congestion information in a collective manner.

It is a schematic block diagram which shows the structure of the vehicle number density observation apparatus which concerns on the Example of this invention. It is a flowchart figure which shows a series of processes until the presentation of the vehicle number density based on the input image and map information in the vehicle number density observation apparatus which concerns on a present Example. It is a figure which shows an example of the input image in the vehicle number density observation apparatus which concerns on a present Example. It is a figure which shows an example of the map information in the vehicle number density observation apparatus which concerns on a present Example. It is a figure which shows the local region division example in the vehicle number density observation apparatus which concerns on a present Example. It is a figure which shows the road direction calculation example in the vehicle number density observation apparatus which concerns on a present Example. It is a figure which shows the example of line segment group extraction in the vehicle number density observation apparatus which concerns on a present Example. It is a figure which shows the example of vehicle line segment extraction in the vehicle number density observation apparatus which concerns on a present Example. It is a figure which shows the example of calculation of the vehicle number density in the vehicle number density observation apparatus which concerns on a present Example. It is a figure which shows the example of a display of the image processing result of the vehicle number density in the vehicle number density observation apparatus which concerns on a present Example.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Vehicle number density observation apparatus 2 Processing apparatus 3 Memory | storage device 4 Keyboard 5 Display apparatus 10 Control part 11 Image input part 12 Road area extraction part 13 Local area division part 14 Road direction calculation part 15 Line segment group extraction part 16 Vehicle line segment extraction part 17 Vehicle Number Density Calculation Unit 18 Vehicle Number Density Display Processing Unit 19 Overlay Processing Unit I Input Image M Map Information

Claims (10)

A vehicle number density observation device for observing the vehicle number density on a road using an input image based on an image taken from the sky and map information,
Means for superimposing the input image and the map information;
Means for extracting only the road area from the superimposed information;
Means for dividing the extracted road area into local areas;
Means for determining the road direction for each divided local area;
Means for extracting line segments from the entire input image;
Means for extracting only the vehicle line segments corresponding to the front or rear surface of the vehicle from the obtained line segments;
Means for determining the density of the vehicle line segment group for each of the divided local regions;
Means for performing image processing to synthesize the obtained vehicle unit density on the corresponding map information;
A presentation means for visually presenting image processing results;
A vehicle number density observation apparatus comprising:   2. The vehicle number density observation apparatus according to claim 1, wherein the image photographed from the sky does not require clearness and resolution to such an extent that a vehicle shape can be easily extracted automatically.   2. The vehicle number density observation apparatus according to claim 1, wherein the means for extracting only the road area extracts only an area to be compared with road area information in the overlapped map information from the image.   2. The vehicle number density observation apparatus according to claim 1, wherein the means for dividing the road area into local areas divides the road area at the intersection, the median strip, and the lane in the overlapped map information.   2. The vehicle number density observation apparatus according to claim 1, wherein the means for obtaining the road direction for each of the divided local areas uses the direction of the boundary line of the road area as the direction in the road area.   The means for extracting a line segment group from the entire input image has a straight line shape among connected components of pixels with high edge strength, has a length corresponding to the length of the vehicle, and is at the boundary of the shadow area. The vehicle number density observation apparatus according to claim 1, wherein non-corresponding ones are extracted.   The means for extracting only the vehicle line segment deletes only the vehicle line segment by deleting a line segment that is unrelated to the front and rear surfaces of the vehicle, for example, a boundary line of a building or a sign on a road. Item 1. The vehicle number density observation apparatus according to Item 1.   2. The number of vehicles according to claim 1, wherein the means for calculating the number of vehicles for each local region obtains the number of vehicles by dividing the number of vehicle line segments in the local region by the area of the local region. Density observation device.   The vehicle number density observation apparatus according to claim 1, wherein the presenting means presents the vehicle number density on the map information as a predetermined observation result. A program for a vehicle number density observation apparatus for observing the number density of vehicles on a road using an input image based on an image taken from the sky and map information,
In vehicle density observation equipment,
A process of superimposing the input image and the map information;
Processing to extract only the road area from the superimposed information;
A process of dividing the extracted road area into local areas;
A process for obtaining a road direction for each divided local area;
Processing to extract line segments from the entire input image;
A process of extracting only the vehicle line segments corresponding to the front or rear surface of the vehicle from the obtained line segments;
Processing for determining the density of the vehicle line segment group for each divided local region;
A process of performing image processing to synthesize the obtained vehicle unit density on the corresponding map information;
Processing for visually presenting image processing results on a display device;
The program for vehicle number density observation apparatuses characterized by performing each process of.