JP4850161B2 - Road map data creation device - Google Patents

Description

  The present invention relates to road map data creation technology, and in particular, analyzes an observation image obtained by photographing a ground surface from an altitude and the sky using a satellite, an aircraft, etc., and in particular, easily extracts a road included in a geographic image. The present invention relates to a road map data creation device for

  In recent years, the use of satellite images and aerial photograph images (hereinafter referred to as “geographic images”) has become widespread, and techniques relating to the creation of maps and the analysis of the ground surface using these images have been developed. Among them, for example, in order to create a road map for car navigation (hereinafter abbreviated as “car navigation”), it is necessary to extract a road area in an image.

  Conventionally, a road region extraction process from a geographic image has been performed by visual inspection. Therefore, there has been a problem that a great human cost is required. Therefore, a method of automating the process of extracting the road area in the image by a computer has been proposed. As an example, in Non-Patent Document 1 below, first, a cover classification process is applied to a geographic image to extract a region having a color tone characteristic like a road. In addition, an edge extraction process and a noise removal process are applied to the geographic image to extract a region having an edge feature like a road. Based on the information of these two areas, the road center line is tracked. Specifically, the user designates one start point for each road on the image obtained by superimposing the two regions. Then, the program can automatically track the road center line from the starting point and extract the center line of one road. Thereafter, the user can extract all road centerlines in the image by designating start points for all roads in the image. By such processing, the road center line can be extracted semi-automatically. Thereafter, road width information is added by visual work, and final road area data is created.

Jun Kumagai, Masafumi Nakagawa, Satoshi, Ryosuke Shibasaki, “Road Extraction from High-Resolution Satellite Images” Proceedings of the 2002 Annual Meeting of the Photogrammetric Society of Japan, 155-158, 2002. J. Canny, “A Computational Approach to Edge Detection” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. PAMI-8, No6, November 1986.

  However, in the method of Non-Patent Document 1, it is necessary to specify a starting point for each road, and this work is complicated and expensive. This method is a process of extracting the center line of the same road a plurality of times when tracking the road center line, and the extracted data may be duplicated. Furthermore, the road information extracted semi-automatically by this method is only the vector data of the road center line, and does not have road width information. In the map information, the width information of the road is highly important information. Therefore, it is necessary to add the width information later based on the road information extracted by this method. In addition, in order to work on a wide range of images in a limited time, generally a plurality of workers are required. In this case, the accuracy of the width is reduced due to the difference in subjective judgment criteria of each worker. There is also a problem that there is a high possibility that variations will occur.

  The present invention has been made in view of such a situation, and road map data creation technology capable of easily and uniformly creating road area data including width information from geographic images such as satellite images and aerial photograph images. The purpose is to provide.

  As a result of diligent research in view of the above problem, the present inventor automatically sets a starting point for image data obtained by superimposing road candidate areas and road candidate edges extracted by the method of Non-Patent Document 1. It was conceived that the above problem can be solved by adding a process to the above, a process of integrating overlapping portions of the extracted road data, and a process of estimating the width of the road.

  According to one aspect of the present invention, a road map data creation device that analyzes a geographic image and creates road map data, wherein a plurality of start points are uniformly distributed in the region of the geographic image extracted from the geographic image data. A road defined by the coordinates at each start point and the coordinates of the node associated with it along a tracking line obtained by following the road from each of the given start points. A road tracking processing unit for extracting vector data, a road integration processing unit for integrating vector data estimated as duplicate data among vector data of the road in each road region in the geographic image, and in the geographic image data A road width estimation processing unit that estimates the width of each road area based on the number of integrated vector data. Data creation device is provided.

  It is preferable that the width estimation processing unit estimates the width as the number of the integrated vector data increases, and increases the width as the vector data of the integrated road is shorter.

  The road integration processing unit searches for the closest node from the extracted nodes of the first road vector and the second road vector to form a node pair, and along the tracking line direction in the node pair It is preferable to search for a node pair whose distance between nodes is within a certain range with respect to the preceding and succeeding nodes, set the searched node pair as a node to be integrated, and generate a new node between the node pairs to be integrated.

  As described above, the geographic image processing apparatus according to the present invention can improve the efficiency of extracting the center line and width of a road from a geographic image such as a satellite image or an aerial photograph image. Further, since the work process that relies on the skill of the worker can be automated, the influence of the difference due to the skill level of the worker can be reduced.

  Hereinafter, a geographic image processing technique according to an embodiment of the present invention will be described in detail with reference to the drawings. 1 to 13, the same reference numerals denote the same configuration, and the basic configuration and operation are the same. Note that the present invention is not limited to the scope described in the present embodiment. A person skilled in the art can make various changes and improvements to the configuration and function of the invention according to the embodiments without departing from the gist of the present invention.

<System configuration>
FIG. 1 is a functional block diagram showing a configuration example of a geographic image processing apparatus according to this embodiment. As shown in FIG. 1, a geographic image processing apparatus includes a processing apparatus 10 composed of a personal computer, a workstation, and the like, and an auxiliary storage device such as a RAM (random access memory) and a magnetic disk storage device used as a main memory. And the input / output device 30.

  The input / output device 30 includes, for example, an input device 31 including a pointing device such as a keyboard and a mouse, a display device 32 such as a liquid crystal display device, and a printer 33. The input device 31 is used for parameter input by a user, activation of a command, and the like, and land parcel data generation processing using a geographic image processed by the geographic image processing device. The display device 32 and the printer 33 are used for presenting geographic images and land parcel data to the user. Note that the display device may include only one of the display device 32 and the printer 33.

  The processing device 10 includes a geographic image processing program 40. As a program module, the geographic image processing program 40 is based on the road candidate area extraction unit 101 that extracts a road-like area (road candidate) based on the color tone information included in the geographic image, and the edge information of the geographic image. The road candidate edge extraction unit 102 for extracting road-like edges, the road tracking processing unit 103 for extracting road centerlines based on road-like areas and road-like edges, and overlapping portions of the extracted road centerlines are integrated. And a width estimation processing unit 105 that estimates the extracted width information of the road.

  The storage device 20 includes a geographic image data storage area 21, a road candidate area data storage area 22, a road candidate edge data storage area 23, and a road area data storage area 24. In the figure, the term “region” is omitted. Of these storage areas, the geographic image data storage area 21 is an area for storing data obtained from an artificial satellite or an aerial photograph image, and is data stored in advance before execution of processing by the geographic image processing apparatus. . On the other hand, the road candidate area data storage area 22, the road candidate edge data storage area 23, and the road area data storage area 24 are generated based on the geographic image data storage area 21 data in the geographic image processing apparatus.

<Outline of geographic image processing>
FIG. 2 is a diagram showing an outline of image processing. FIG. 2A shows an image based on geographic image data stored in the storage device 20 in advance. Cover classification processing is performed on the geographic image data, and a road-like area is extracted based on the result, thereby obtaining a road candidate area image shown in FIG. This road candidate area image is stored in the road candidate area data storage area 22. Further, an edge extraction process is performed on the geographic image data, and noise that is not likely to be a road is removed from the extracted edges, whereby a road candidate edge image shown in FIG. 2C is obtained.

  This road candidate edge image holds data representing the edge direction at each pixel. This data is stored in the road candidate edge data storage area 23 as road candidate edge data. Then, based on the road candidate area data and the road candidate edge data, a road tracking process is performed to extract the center line of the road. In the road tracking process, the road is automatically tracked from the given starting point. When the road is tracked from each starting point, an image displaying the road tracking line shown in FIG. 2D is obtained. This road tracking line (white thin line in the figure) is vector data having coordinates at each node and node information connected thereto. This vector data is stored in the road area data storage area 24 as road area data. However, a noise-like display exists in the upper right area. There may be a plurality of road extraction lines in the road area data on the same road, and in this case, the estimated data becomes redundant. Therefore, noise is removed by performing processing (road integration processing) for integrating duplicate data.

  FIG.2 (e) is a figure which shows the result of having integrated the road data in FIG.2 (d). The integrated vector data is overwritten and stored in the road area data storage area 24. In the integrated data, it can be seen that the noise is almost removed and represented as a clean white line. Next, a width estimation process is performed. The width of the road is estimated from the road vector length of each road in the road area data and the number of integrated vectors. FIG. 2F is a diagram illustrating a result of estimating the width of the vector data of each road in the road data. A bright area is an area extracted as a road, and the width is shown. The width data is added to the road area data storage unit 24.

<Details of geographic image processing>
Details of each process in the geographic image process will be described below. In the geographic image processing apparatus according to the present embodiment, when the geographic image processing program 40 is started, a road candidate area extraction unit 101, a road candidate edge extraction unit 102, a road tracking unit 103, a road integration processing unit 104, and a width estimation process. Units 105 are sequentially activated. Hereinafter, each process in each processing unit will be described in detail.

(1) Road Candidate Area Extraction Processing FIG. 3A is a flowchart showing the flow of processing by the road candidate area extraction unit 101 of the geographic image processing program 40. In FIG. 3 (a), the <Same as below> road candidate area extraction unit 101 shown in FIG. 1 reads geographic image data from the geographic image data storage area 21 of the storage device 20 (S301), and covers classification (the above non-patent document). 1) (S302). Cover classification is a method of classifying each pixel included in a geographic image into various predetermined object classes (roads, buildings of different colors, trees, water areas, soil, etc.). In recent years, with the development of remote sensing technology, high-resolution satellite image data of 1 m or less has become available. Unlike conventional low-resolution satellite images, surface structures such as artificial structures, trees, and roads can be recognized to some extent even by cover classification using a computer. In addition, aerial photograph images have similar characteristics.

  FIG. 3B is a diagram showing the result of the cover classification process for the image of FIG. In this image, it is classified into several classes of surface objects such as roads, buildings of various colors, and trees. However, in reality, an area classified as “road” does not always accurately represent an actual road, and usually there are cases where an extraction omission or an erroneous extraction occurs. Therefore, based on the covering classification result, road candidate region extraction processing is performed in order to compensate for the region where extraction has been omitted and to reduce the influence of erroneous extraction (S303). When extracting candidate road areas, it is easy to misidentify classes that are easily misidentified as roads other than road classes (for example, classes of buildings with white or black color characteristics) from the coverage classification results. This is extracted in consideration of the weight. In addition, classes that are unlikely to be mistaken for roads (for example, classes such as trees and soil) are excluded from this processing target. By adding such processing, it is possible to extract an area (road candidate area) that seems to be a road from the cover classification result image. FIG. 3C is a diagram showing a road candidate area, and a brightly shown area is a road candidate area. The data specifying the road candidate area shown in FIG. 3C is output and stored in the road candidate area data storage area 22 as road candidate area data (S304).

(2) Road Candidate Edge Extraction Processing FIG. 4A is a flowchart showing the flow of processing by the road candidate edge extraction unit 102 of the geographic image processing program 40 shown in FIG. In FIG. 4A, the road candidate edge extraction unit 102 reads map image data from the geographic image data storage area 21 of the storage device 20 (S401), and performs edge extraction (S402). Although there are various methods for edge extraction processing in an image, the Canny method (see Non-Patent Document 2 above) is particularly suitable. FIG. 4B is an image obtained as a result of performing edge extraction on the geographic image of FIG. 2A by the Canny method. This image contains road edges, but may also contain many noises at the same time. Therefore, next, noise other than roads is removed from this image (S403). When noise is removed, short edges or edges whose direction is not changed like a road are removed (see Non-Patent Document 1 above). FIG. 4C is an image obtained as a result of performing noise removal processing on the image of FIG. The edge data on each pixel after noise removal is stored in the road candidate edge data storage area 23 as road candidate edge data (S404).

(3) Road Tracking Process FIG. 5 is a flowchart showing the flow of processing in the road tracking processing unit 103 of the geographic image processing program 40. In FIG. 5, the road tracking processing unit 103 shown in FIG. 1 (the same applies hereinafter) stores the respective data from the road candidate area data storage area 21 and the road candidate edge data storage area 22 of the storage device 20 (the road in FIG. 3C). The candidate area data and the road candidate edge data in FIG. 4C are read, and the two data are superimposed on the image (S501). FIG. 6A is a diagram showing a state in which these two data are superimposed. The road tracking processing unit 103 (FIG. 1) performs road tracking in this figure and extracts road centerline data. When performing road tracking, first, a starting point for tracking is set (S502). It is preferable to set a large number of starting points for the entire image. For example, as shown in FIG. 6B, it is preferable to set a large number of start points that are parallel to the x axis and the y axis, which are coordinate axes orthogonal to each other in the image, at equal intervals. According to such a method, there is an advantage that the user does not need to set a starting point.

  Next, the road is tracked from each starting point. The road tracking process is preferably performed by a line template matching method (see Non-Patent Document 1 above). First, one unprocessed start point is selected (S503), and the start direction is determined (S504). When determining the start direction, as shown in FIG. 6C, a rectangular template T1 is superimposed on the image, and road candidate area data and road candidate edge data are referred to. Then, the angle of the template T1 is changed and the road-likeness calculated from both data at each angle is measured, and a more road-like direction is set as the start direction. In the case of FIG. 6C, the upper left direction is more likely to be a road (can be determined by the fact that there are many bright pixels in the road candidate area data and road candidate edge data), so the upper left direction is taken as the start direction. Determine (starting direction is the direction of the arrow in the figure). After determining the start direction, the extension direction is searched (S505). Even when the extension direction is determined, the direction of the road is changed while changing the angle of the line template (the process of making the extension direction is continued (the bright rectangle in FIG. 6 (d) and the rectangle with different angles indicated by the broken line)). .

  However, unlike the determination of the start direction, it is preferable to narrow the angle range to be changed when determining the extension direction. This is because roads generally do not bend so rapidly except for special cases such as intersections and mountainous areas and often draw relatively gentle curves. When the extension direction is determined (the direction where there are many bright pixels in the road candidate edge data), the road candidate edge data is extended by the length of the line template, and the extension destination point is set as a node (starting point) constituting the road vector. ).

  Here, when determining the extension direction by changing the angle of the line template, if it is determined that the road-likeness is low in any direction, the extension process is stopped, the road tracking is ended, and the opposite of S507 in the flowchart is performed. Side processing is performed (S506). If the extension direction is determined, the extension direction is determined again at the extension destination (S505). This process is performed until the extension direction cannot be determined, that is, until the road is interrupted. Such road tracking is also performed on the side opposite to the start direction determined in S504. S507 is processing for determining whether or not the other side has been tracked. After completing the tracking of the road on both sides with respect to the start point set in S503, the connection data of the line segment generated by combining the nodes set at the time of tracking is set as one road area data (FIG. 6). L1 of (e). The processes from S503 to S507 are performed for each start point set in S502. S508 is the determination process. FIG. 6F shows the result after the road tracking process. As shown in FIG. 6F, it can be seen that a large number of thin lines are extracted in order to track the road based on a large number of starting points. Each road vector data is stored in the storage device 20 <FIG. 1> in the format as shown in FIG. However, since the “integrated number” and “width” items are not obtained at this stage, data of these items is not stored. Since the item “road vector length” is calculated in the width estimation process described later, data is not stored at this stage. However, since data can be calculated at this stage, data may be stored.

(4) Road Integration Processing FIG. 7 is a flowchart showing the flow of processing in the road tracking processing unit 103 shown in FIG. In FIG. 7, the road integration processing unit 104 reads road area data from the road area data recording area 24 of the storage device 20 (S701), and reads only one road vector from the road area data (S702). For each road vector, integration processing is performed with neighboring road vectors. First, it is checked whether or not there is a road vector (B) whose distance and direction are close in the vicinity of the road vector (A) of interest (S703). If no road vector having a close distance and direction is found nearby, the process returns to S702, and another road vector is read. If a road vector with a close distance and direction is found nearby, the process proceeds to S705, where the two road vectors (A) and the road vector (B) are estimated as data to be integrated, and these are integrated. FIGS. 8B and 8C are step diagrams showing how the two road vectors (A) and (B) extracted for the road as shown in FIG. 8A are integrated. . When the road vectors (A) and (B) are integrated, first, the closest node is searched from the nodes of the two road vectors, and this node pair P1, P2,. In this reference node pair P1, P2,..., The previous and next nodes are examined, and a node pair in which the distance between the nodes in each node pair P1, P2,.

  The node pair searched here is set as a node to be integrated. At this time, if it is determined that the distance between the nodes is equal to or greater than a certain distance and there are no adjacent nodes, the search process is terminated. Next, the node after integration is set at an intermediate position of the node pair to be integrated. In FIG. 8B, since there are six node pairs to be integrated, six new nodes PA1,... Are set in the middle of each node pair. These new nodes PA1,... Are connected in the closest order, and a connected line segment generated by combining them is set as a new road vector. At this time, a new node coordinate may be obtained by the integration ratio. In FIG. 8C, since the integration ratio is 2: 5, the node coordinates indicate the integration number as shown in the drawing drawn below. It will be close to the larger one.

  Further, in a portion where there is no adjacent node and is cut in the previous processing, the node of the cut portion and the node of the new road vector are combined. At this time, the node of the branched portion is shared by the three road vectors. Further, at the time of integration, the number of road vectors integrated so far (number of integration) is stored. This information on the number of integrations can be used when the width is estimated in the width estimation process described later. After the road vectors (A) and (B) are integrated, the road area data is updated (S706). At this time, since the road ID may increase or decrease due to the integration, the road ID is reassigned. The processing from S702 to S706 is performed for all road vectors (S707). Finally, it is checked whether or not the road area data has been updated as a result of the processing from S702 to S707 (S708). If at least one road vector has been updated in the processes from S702 to S707, the process returns to S702 and the process is recursively repeated. If there is no update, the process ends.

  FIG. 9 is a schematic diagram showing road integration processing. There are four road vectors (ID: 0001 to 0004) on the bifurcated road, and as a result of applying the road integration process, one road vector (ID: 0001) with an integrated number of four is integrated. A state in which two road vectors (ID: 0002, 0003) of Formula 2 are generated is shown.

(5) Width Estimation Processing FIG. 11 is a flowchart showing the flow of processing in the width estimation processing unit 105 (see FIG. 1) of the geographic image processing program 40. In FIG. 11, the width estimation processing unit 105 reads road area data from the road area data storage area 24 of the storage device 20 (S1101). Next, the processing after S1102 is performed on each road vector. First, for each road vector, the road vector length (total length of short vectors between nodes) is calculated (S1102). Next, the width is estimated from the road vector length of each road vector and the number of integrations. In the present embodiment, when the road vector is extracted by the road tracking processing unit 103, the start point is set uniformly for the entire image, so that the larger the road area, the more road vectors are extracted. . Therefore, when the overlapping road vectors are integrated in the road integration processing unit 104, the number of road vectors to be integrated increases as the area of the road increases. Since the area can be estimated from the length and width of the road, the width can be estimated from the road vector length and the number of road vectors integrated. An example of the approximate width is the number of road vectors integrated / the length of the road vector (ratio). The width is estimated by multiplying the reference width by the above ratio. More specifically, it can be obtained by the following equation (1).

(Number of integrated road vectors / length of road vector) x a (proportional constant) (1)
The proportionality constant may be obtained empirically, or may be determined in advance assuming that the same country or region is the same. Later, it may be corrected by comparison with the actually measured value.

  FIG. 12 is a diagram illustrating how road vectors and widths are extracted according to the present embodiment. FIG. 12A shows an image of a Y-shaped road and a start point set thereon. FIG. 12B is an image obtained as a result of applying the road tracking process to the image of FIG. Since the road is tracked from a plurality of starting points on the road, a plurality of road vectors exist on the same road. FIG. 12C is a result image obtained by applying road integration processing to the result of FIG. The road integration process shows a state in which a plurality of road vectors existing on the same road are integrated. FIG.12 (d) is a result image which applied the width | variety estimation process with respect to the result of FIG.12 (c). This shows the result of estimating the width based on the road vector length data and the integrated number data.

  FIG. 13 is a diagram corresponding to FIG. 10, and illustrates the final processing result of the road area data 24. Each data is stored for each road vector ID. In the road ID00001, the integration number is 60 and the road vector length is 150. In the road ID00002, the integration number is 40 and the road vector length is 100. In S1103, for example, the width is approximated by the integration number / vector length. More specifically, the width is calculated as follows by the above equation (1).

  That is, in the above formula (1), if a = 125 (this value is arbitrary), the width of the road of ID00001 is (60/150) × 125 = 50, and the width of the road of ID00002 is (40/100) X125 = 50, both 50.

<Summary>
As described above, the embodiment of the present invention relates to a road map data creation system that analyzes geographic image data obtained by photographing a ground surface and easily creates road area data on the ground surface. The system includes a road candidate area extraction unit (module) that extracts a road candidate area from geographic image data based on color tone information, and a road candidate based on edge information from geographic image data. A road candidate edge extracting unit (module) for extracting a region, a road tracking processing unit (module) for tracking the road region and extracting road vector information, and a road integration processing unit for integrating overlapping road vectors ( Module) and a width estimation processing unit (module) for estimating road width information. This makes it possible to automate the extraction of road areas that do not include overlapping width information.

  The present invention can also be realized by a program code of software that implements the functions of the present embodiment. In this case, a storage medium in which the program code is recorded is provided to the system or apparatus, and the computer (or CPU or MPU) of the system or apparatus reads the program code stored in the storage medium. In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiments, and the program code itself and the storage medium storing the program code constitute the present invention. Examples of the storage medium for supplying such program code include a Propi (registered trademark) disk, CD-ROM, DVD-ROM, hard disk, optical disk, magneto-optical disk, CD-R, magnetic tape, and nonvolatile medium. A memory card, ROM, or the like is used.

  Also, based on the instruction of the program code, an OS (operating system) running on the computer performs part or all of the actual processing, and the functions of the above-described embodiments are realized by the processing. May be. Furthermore, after the program code read from the storage medium is written in the memory on the computer, the CPU of the computer or the like performs part or all of the actual processing based on the instruction of the program code. You may make it implement | achieve the function of embodiment mentioned above.

  Also, software program codes for realizing the functions of the embodiments are stored in a storage means such as a disk or memory or a storage medium such as a CD-RW or CD-R, and a computer (or CPU or MPU) of the system or apparatus thereof. However, it may be achieved by reading and executing the program code stored in the storage means or the storage medium.

  The present invention can be used in a road map data creation device.

It is a functional block diagram which shows the example of 1 structure of the geographic image processing apparatus by one embodiment of this invention. It is a figure which illustrates the outline | summary of the image processing performed by the geographic image processing apparatus according to this embodiment. It is a flowchart figure which shows the flow of a process of a road candidate area | region extraction part. It is a flowchart figure which shows the flow of a process of a road candidate edge extraction part. It is a flowchart figure which shows the flow of a process of a road tracking process part. It is a figure which illustrates the outline | summary of the process of a road tracking process part. It is a flowchart figure which shows the flow of a process of a road integrated process part. It is a figure which illustrates the outline | summary of the process of a road integrated process part. It is a figure which illustrates the result of the process of a road integrated process part. It is a figure which illustrates the contents of road field data. It is a flowchart figure which shows the flow of a process of the width | variety estimation process part. It is a figure which illustrates the outline | summary of the process of a width estimation process part. It is a figure which illustrates a final processing result.

Explanation of symbols

10 processing device 20 storage device 21 geographic image data storage region 22 road candidate region data storage region 23 road candidate edge data storage region 24 road region data storage region 30 input / output device 31 input device 32 display device 33 printer 40 geographic image processing program 101 Road candidate region extraction unit 102 Road candidate edge extraction unit 103 Road tracking processing unit 104 Road integration processing unit 105 Width estimation processing unit

Claims (3)

A road map data creation device that analyzes geographic images and creates road map data,
A plurality of starting points are uniformly provided in the region of the geographic image extracted from the geographic image data, and each of the plurality of starting points is traced along a tracking line obtained by tracking the road from each of the given starting points. A road tracking processing unit that extracts vector data of a road defined by the coordinates at the start point and the coordinates of the node connected to the start point;
In each road area in the geographic image, a road integration processing unit that integrates vector data estimated as duplicate data among vector data of the road;
A width estimation processing unit that estimates the width of each road area in the geographic image data based on the number of integrated vector data;
A road map data creation device comprising:   The width estimation processing unit estimates the width so that the width increases as the number of the integrated vector data increases, and the width increases as the length of the vector data of the integrated road decreases. Item 2. The road map data creation device according to Item 1.   The road integration processing unit searches for the closest node from the extracted nodes of the first road vector and the second road vector to form a node pair, and along the tracking line direction in the node pair A node pair having a distance between nodes within a certain range is searched for the preceding and following nodes, the searched node pair is set as a node to be integrated, and a new node is generated between the node pairs to be integrated. Item 3. A road map data creation device according to item 1 or 2.

Images