JP5472137B2 - Boundary detection device and boundary detection program - Google Patents

Description

  The present invention relates to a boundary detection device that detects a boundary of a travel region in which a vehicle travels, and a boundary detection program.

  As the boundary detection device, a device that detects a boundary of a travel region by performing image processing on a captured image obtained by capturing the traveling direction of a vehicle is widely known. In such an apparatus, a method of extracting a part where the luminance greatly changes between adjacent pixels as an edge component (edge point) and detecting a boundary of the traveling region from the edge component is usually adopted. It is known that a road model is generated based on map data and an edge component that deviates from the road model is excluded to improve the detection accuracy of the boundary of the travel region (for example, see Patent Document 1).

JP 2010-170488 A

  However, in the boundary detection apparatus, a road model is generated from map data, so a database for storing map data is required. Further, when the map data and the actual road shape are different, there is a possibility that the edge component indicating the boundary of the travel area is excluded.

  Accordingly, in view of such problems, a boundary detection apparatus for detecting a boundary of a travel region in which a vehicle travels can accurately detect a boundary of the travel region with a simple configuration that does not require a map data database. This is an object of the present invention.

  In the boundary detection device having the first configuration configured to achieve the above object, the edge direction calculation means includes a vertical direction of the captured image for each of a large number of pixels constituting the captured image captured around the host vehicle. By detecting the magnitude of the luminance change between adjacent pixels and the magnitude of the luminance change between adjacent pixels in the horizontal direction, and comparing the magnitude of the luminance change in each direction, the direction in which the luminance changes most greatly is determined. Calculate the edge direction to represent. Further, the edge component extraction means extracts an edge component representing an area where the luminance change between adjacent pixels is equal to or greater than the luminance reference value for each of a large number of pixels. Then, the angle calculation means calculates the angle formed by the edge direction with respect to the direction of the virtual reference straight line representing the straight line connecting the position of the infinity point in the captured image and the position of the edge component for each edge component. The first weighting setting unit sets the weighting for each edge component so that the weighting increases as the angle formed becomes closer to a right angle. Further, the edge detection means detects the edge of the boundary of the traveling region by performing image processing in consideration of the weighting set for each edge component.

  According to such a boundary detection device, weighting is set for each edge component according to the angle formed by the edge direction and the virtual reference line, and image processing is performed in consideration of this weighting. The influence of the edge component having the direction in the edge direction that is likely to be the boundary of the traveling region can be increased. Therefore, it is possible to accurately detect the boundary of the travel region with a simple configuration.

  In the present invention, when extracting the edge component, the luminance change in any direction such as the vertical direction, the horizontal direction, or the edge direction of the captured image may be compared with the luminance reference value. In addition, as the image processing of the present invention, a known process for obtaining a curve (straight line) serving as a boundary of the traveling region, such as a well-known voting process such as a Hough transform or a least square method, can be employed.

  By the way, in the boundary detection device, as in the second configuration, the boundary detection device is set to repeatedly detect the boundary of the traveling region, and this traveling region is detected each time the boundary of the traveling region is detected. An infinity point updating means for detecting the position of a new infinity point in the captured image in accordance with the position of the boundary and updating the previous infinity point position to a new infinity point position, and an angle calculation means May calculate the angle formed using the position of the new infinity point.

  According to such a boundary detection device, the position of the infinity point can be fed back and used, so that even when the vehicle inclination changes with the change in the amount of passengers or luggage, it is appropriate It is possible to set the position of the infinity point in the captured image. The configuration of the present invention is effective when it is possible to determine that the boundary of the travel region is substantially a straight line.

  Further, in the boundary detection device, as in the third configuration, the road model for determination indicating the shape of the boundary of the travel region predicted in the captured image based on the shape of the boundary of the travel region detected previously. A determination road model generation means for generating a virtual reference straight line in a tangential direction at a portion corresponding to the position of the edge component of the determination road model for each position of the edge component. The angle formed by the edge direction with respect to the direction of the virtual reference straight line may be calculated.

  According to such a boundary detection device, even when the road on which the vehicle travels is not a straight line, the edge component can be appropriately weighted according to the shape of the travel region. In addition, when combining the second configuration and the third configuration, based on the previously detected shape of the boundary of the travel region, determine whether the boundary of the travel region is a straight line or a curve, If it is a straight line, the second configuration may be adopted, and if it is a curve, the third configuration may be adopted.

  Further, in the boundary detection device, as in the fourth configuration, for each of the edge component positions, a part of the parameters that pass through the position of the edge component and constitute the determination road model is selected. Similar road model generation means for generating similar road models that differ only in the angle calculation means for each position of the edge component, the virtual reference in the tangential direction in the part corresponding to the position of the edge component of the similar road model A straight line may be set, and the angle formed by the edge direction with respect to the direction of the virtual reference straight line may be calculated.

  According to such a boundary detection device, since the virtual reference straight line that passes through the position of the edge component is set, the virtual reference straight line when it is assumed that the boundary of the traveling region exists at the position of the edge component is appropriately set. Can do.

  Further, in the boundary detection device, as in the fifth configuration, for parameters that are different between the similar road model and the determination road model, the value of this parameter in the similar road model and the value of this parameter in the determination road model And a second weight setting means for setting the weight for each edge component so that the weight increases as the difference decreases.

  According to such a boundary detection device, it is possible to set a large weight to the edge component constituting the similar road model close to the determination road model, so that the boundary of the travel region close to the determination road model is detected. It can be made easier.

  Next, a sixth aspect of the boundary detection program configured to achieve the above object is a computer as each means constituting the boundary detection device according to any one of claims 1 to 5. It is a program for making it function.

  According to such a boundary detection program, at least the same effect as that of the enhancement detection apparatus according to claim 1 can be obtained.

It is a block diagram which shows the structure of the boundary detection system 1 mounted in the vehicle. It is the flowchart (a) which shows a boundary detection process, and the flowchart (b) which shows a road model setting process. It is a flowchart which shows a white line candidate extraction process (a), and an edge intensity | strength and direction calculation process (b). It is explanatory drawing which shows edge strength and edge direction. It is the flowchart (a) which shows a weighting process, and the graph (b) and (c) which show the value of a specific weight. It is explanatory drawing which shows a local infinity point.

Embodiments according to the present invention will be described below with reference to the drawings.
[Configuration of this embodiment]
FIG. 1 is a block diagram illustrating a configuration of a boundary detection system 1 mounted on a vehicle. FIG. 1A is an overall configuration, and FIG. 1B is a detailed configuration of a lane boundary detection device as a main part. As shown in FIG. 1A, the boundary detection system 1 includes a lane boundary detection device 10 (boundary detection device) and a vehicle control ECU 20.

  The lane boundary detection device 10 has a function of generating a boundary parameter indicating a position or the like that becomes a boundary (hereinafter referred to as “lane boundary”) of a lane (traveling area) in which the host vehicle is traveling. The vehicle control ECU 20 is connected to the lane boundary detection device 10 via an in-vehicle LAN, and deviates from the lane in which the host vehicle is traveling based on boundary parameters acquired from the lane boundary detection device 10 through the in-vehicle LAN. This function has a function of executing various processes for obtaining the possibility and notifying the driver of the possibility.

  As shown in FIG. 1B, the lane boundary detection device 10 includes a CCD camera 11, an analog / digital converter (ADC) 12, an image memory 13, a CPU 14, a ROM 15, a RAM 16, and a communication IC 17. It has. The CCD camera 11 is arranged so as to image the road surface in the traveling direction of the vehicle (front of the vehicle).

  The ADC 12 converts an analog image signal output from the CCD camera 11 into digital image data. The image memory 13 stores the image data obtained by the ADC 12.

  The CPU 14 executes processing for generating boundary parameters based on the image data stored in the image memory 13. The ROM 15 stores a program for processing executed by the CPU 14, and the RAM 16 functions as a work area for the CPU 14.

  The communication IC 17 outputs the processing result in the CPU 14 to the outside via the in-vehicle LAN. Like the lane boundary detection device 10, the vehicle control ECU 20 includes a CPU, a ROM, a RAM, and a communication IC. In addition, the vehicle control ECU 20 receives a detection signal from a sensor directly connected to the ECU 20, and outputs a control signal for a control target. An IO port or the like is provided.

  Here, at least a speaker that generates an alarm sound is connected as a control target. The vehicle control ECU 20 may be configured to function as a device for preventing a traffic accident such as a steering torque control device that controls the steering torque so that the vehicle does not depart from the travel lane.

[Process of this embodiment]
Next, boundary detection processing (contents of processing by the boundary detection program) executed by the CPU 14 will be described with reference to the flowcharts shown in FIG. This process is a process that is started, for example, when a vehicle such as an ignition switch is turned on, and then repeatedly performed at regular intervals (for example, every 100 ms).

  FIG. 2A is a flowchart showing the boundary detection process, and FIG. 2B is a flowchart showing the road model setting process in the boundary detection process. FIG. 3A is a flowchart showing white line candidate extraction processing in the boundary detection processing, and FIG. 3B is a flowchart showing edge strength / direction calculation processing in white line candidate extraction processing. FIG. 5A is a flowchart showing the weighting process in the white line candidate extraction process.

  In the boundary detection process, as shown in FIG. 2A, first, a road model setting process is performed (S110). The road model setting process estimates the shape (curvature, etc.) of the road on which the vehicle travels using the white line position (boundary of the travel region) detected in the previous boundary detection process and the motion state of the vehicle. This is processing for setting the estimated position of the white line in the captured image as a road model.

  Specifically, as shown in FIG. 2B, the road model setting process first acquires the position of the infinity point detected last time (S210). Then, it is determined whether or not the position of the infinity point has been acquired (S220).

  Immediately after the boundary detection process is started (that is, when the boundary detection process is started for the first time), there is no information on the position of the infinity point detected before the previous time. In the case, an affirmative determination is made. If the position of the point at infinity has been acquired (S220: YES), the process proceeds to S240 described later.

  If the position of the point at infinity has not been acquired (S220: NO), an estimated road model is set (S230). For example, as shown in FIG. 4A, the estimated road model set here includes a position of an infinite point (FOE) and two linear white lines extending toward the infinite point. The

  Subsequently, the vehicle speed and yaw rate of the host vehicle are acquired from a well-known vehicle speed sensor (not shown) and yaw rate sensor (not shown) connected via the in-vehicle LAN (S240), and zero point correction of the yaw rate is performed (S250). . For the zero point correction of the yaw rate, a known process as disclosed in, for example, Japanese Patent Laid-Open No. 9-49875 may be performed.

  Next, the curvature of the road on which the vehicle travels is calculated using the detection results of the vehicle speed sensor and the yaw rate sensor mounted on the vehicle, and a road model for determination is generated and set (S260: road model for determination). Generating means).

  Here, the determination road model is generated by correcting the estimated road model set in the process of S230 or the previous boundary detection process (the process of S150 described later). Specifically, the estimated road model includes information such as the lane width (using the default value or the previously detected value), the shape of the white line, the position of the infinity point, and the like. However, since the estimated road model shows the past (or default value) road shape, this process uses the estimated road model by using information such as the vehicle speed, turning angular velocity, curve curvature, etc. The model is estimated to be obtained from the captured image.

  For example, it predicts that the vehicle will move forward according to the vehicle speed, and generates a white line shape that is estimated to be imaged at the predicted position. When the curvature of the curve included in the estimated road model is different from the curvature of the curve by the yaw rate sensor, the shape of the white line is corrected according to the curvature of the curve by the yaw rate sensor.

  At this time, as the curvature of the curve of the road model for determination, the curvature of the curve by the yaw rate sensor may be employed as it is, or a weighted average with the curvature of the curve of the estimated road model may be employed. When such processing ends, the road model setting processing ends.

  Subsequently, returning to FIG. 2A, white line candidate extraction processing is performed (S120). The white line candidate extraction process is a process of extracting a white line candidate (a boundary candidate that may be a boundary of the travel region) from the captured image.

  Specifically, as shown in FIG. 3A, first, an image captured by the camera 11 is acquired from the CCD camera 11 (S310), and edge direction / direction calculation processing is performed (S320: edge direction calculation means). . This process is performed for each selected pixel by sequentially selecting each pixel constituting the captured image. For example, the processing is performed sequentially from the pixel in the upper left corner to the pixel in the upper right corner, and then performed sequentially from the pixel in the lower left corner to the pixel in the right corner.

  In the edge direction / direction calculation processing, as shown in FIG. 3B, first, the vertical edge strength of the captured image is detected for each pixel (S410), and the horizontal edge strength of the captured image is detected (S410). S420). Here, the edge strength means a difference in brightness between adjacent pixels.

The edge strength will be described in detail with reference to FIG. FIG. 4 is an explanatory diagram showing edge direction / direction calculation processing.
For example, when the brightness of each pixel on the line A set in the horizontal direction is detected in a captured image obtained by capturing a linear white line as shown in FIG. 4A, the horizontal position and brightness on the line A are detected. The relationship as shown in FIG. 4B is obtained. If the difference between the brightness of a certain pixel and the brightness of the adjacent pixel on the left side is the amount of change, the relationship shown in FIG. 4C is obtained between the horizontal position on the line A and the amount of change. It is done.

  Similarly, when the brightness of each pixel on the line B set in the vertical direction is detected, the relationship shown in FIG. 4D is obtained between the vertical position on the line B and the brightness. If the difference between the brightness of a certain pixel and the brightness of the adjacent pixel on the lower side is the amount of change, the relationship shown in FIG. 4E is obtained between the vertical position on the line B and the amount of change. It is done. That is, it can be said that the edge strength increases as the difference in brightness between adjacent pixels in each direction (vertical direction or horizontal direction) increases.

  Subsequently, returning to FIG. 3B, the edge direction is calculated based on the edge strengths in the vertical and horizontal directions (S430). Here, the edge direction indicates a direction in which the brightness changes most between certain pixels (a direction perpendicular to a tangent to the white line).

  For example, when the white line is arranged so as to be directly beside the captured image as in a pedestrian crossing, the edge strength in the vertical direction increases, but the edge strength in the horizontal direction becomes almost zero. Further, as shown in FIG. 4A, when the white line is arranged at an angle of about 45 degrees in the captured image, the edge strengths in the vertical direction and the horizontal direction are substantially equal. By using such characteristics, the edge direction can be specified by the balance of the edge strength in the vertical direction and the horizontal direction.

  When such processing ends, the edge strength / direction calculation processing ends. Subsequently, returning to FIG. 3A, an edge component included in the captured image is detected (S330: edge extraction means).

  An edge component represents a region where there is a luminance change (edge intensity) that is equal to or greater than a luminance reference value between adjacent pixels among a large number of pixels constituting a captured image, and is generally also referred to as an edge point. As to whether or not there is a luminance change equal to or higher than the luminance reference value, the luminance change (edge strength) in the vertical direction or the horizontal direction may be compared with the luminance reference value, or the luminance change in the edge direction (for example, vertical The root mean square of the edge strength in the direction and the edge strength in the horizontal direction) may be compared with the luminance reference value.

  Subsequently, a weighting process is performed (S350). The weighting process is a process of setting a weight for each edge component according to the validity of the position and edge direction of the edge component. Specifically, as shown in FIG. 5A, first, a similar road model that passes through the position of the edge component and is similar to the determination road model is calculated (S510: similar road model generation means).

  For example, the similar road model is a model in which only the curvature of the curve (the radius of the curve) is different from the determination road model. Subsequently, the radius of the curve of the similar road model is compared with the radius of the curve of the road model for determination, and weighting according to these differences is performed (S520: second weight setting means). Specifically, for example, as shown in FIG. 5B, the coefficient (weighting) is decreased as the difference in the radius of the curve increases.

  Next, a local infinity point (local FOE) of the similar road model is calculated (S530). In this process, as shown in FIG. 6, for each position of the edge component for which a local infinity point is to be calculated, a similar road model is used at the position of the edge component and the part located at the same distance from this position. A tangent line is drawn to find an infinite point for each edge component position.

  Then, an angle formed by a straight line connecting the position of the edge component and a local infinity point with respect to the edge component and the edge direction is calculated, and weighting of the edge component is set according to the formed angle (S540). : Angle calculation means, first weight setting means). In this process, for example, as shown in FIG. 5C, the weight is set for each edge component so that the weight becomes larger as the angle formed becomes closer to the right angle, and the coefficient set in the process of S520 is performed. The product of (see FIG. 5B) and the coefficient set in S540 (see FIG. 5C) is set as a weight for the edge component.

  When such processing is finished, the weighting processing is finished. Subsequently, returning to FIG. 3A, a candidate for a curve (including a straight line) serving as a boundary of the traveling region is extracted as a white line candidate by performing well-known Hough transform based on the arrangement of each edge component (S360: End detection means). However, when this Hough transform is performed, weighting is taken into consideration, and an edge component with a larger weight is more likely to be selected as a white line candidate. When such processing ends, the white line candidate extraction processing ends.

  Then, returning to FIG. 2A, the most probable white line candidate is set as a white line indicating the boundary of the travel region and recorded in a memory such as the RAM 16 (S130). Subsequently, in the captured image, an intersection point when the white line on both sides of the host vehicle is extended is set as an infinite point, and is overwritten and recorded in a memory such as the RAM 16 (S140: infinite point updating means).

  Then, the shape of the white line is overwritten and recorded in the memory such as the RAM 16 as an estimated road model (S150), and the boundary detection process ends. Note that the estimated road model including the position of the white line and the position of the infinity point recorded in the processes of S130 to S150 is used in the next and subsequent boundary detection processes.

[Effects of this embodiment]
In the lane boundary detection device 10 described in detail above, the CPU 14 has a large luminance change between pixels adjacent to each other in the vertical direction of the captured image for each of a large number of pixels constituting the captured image captured around the host vehicle. And the magnitude of the luminance change between adjacent pixels in the horizontal direction, the edge component representing the area where the luminance change between adjacent pixels is equal to or greater than the luminance reference value is extracted, and the luminance change in each direction is detected for each edge component. By comparing the sizes, the edge direction representing the direction in which the luminance changes the largest is calculated. Then, the CPU 14 calculates the angle formed by the edge direction with respect to the direction of the predetermined virtual reference straight line for each edge component, and weights each edge component so that the weight increases as the formed angle approaches a right angle. Set. Further, the CPU 14 performs image processing in consideration of the weighting set for each edge component, thereby detecting the end of the boundary of the traveling region.

  According to such a lane boundary detection device 10, weighting is set for each edge component according to the angle formed by the edge direction and the virtual reference line, and image processing is performed in consideration of this weighting. In addition, it is possible to increase the influence of the edge component having the direction of the edge direction that is highly likely to be the boundary of the traveling region. Therefore, it is possible to easily detect the boundary of the travel region with a simple configuration and with high accuracy.

  Further, in the lane boundary detection device 10, the CPU 14 determines a road model for determination that indicates the shape of the boundary of the traveling region that is expected in the captured image obtained this time, based on the shape of the boundary of the traveling region detected previously. Generate and set a virtual reference line in the tangential direction at the part corresponding to the position of the edge component in the judgment road model for each position of the edge component, and calculate the angle formed by the edge direction with respect to the direction of this virtual reference line To do.

According to such a lane boundary detection device 10, even when the road on which the vehicle travels is not a straight line, the edge component can be appropriately weighted according to the shape of the travel region.
Further, in the lane boundary detection device 10, the CPU 14, for each edge component position, passes a similar road model that passes through the position of the edge component and differs only in some of the parameters that constitute the determination road model. For each position of the edge component, a virtual reference line is set in the tangential direction at the part corresponding to the position of the edge component in the similar road model, and the angle formed by the edge direction with respect to the direction of the virtual reference line is calculated. .

  According to the lane boundary detection device 10 as described above, the virtual reference straight line that passes through the position of the edge component is set, so that the virtual reference straight line is appropriately set when it is assumed that the boundary of the traveling region exists at the position of the edge component. can do.

  In the lane boundary detection device 10, the CPU 14 compares the value of this parameter in the similar road model with the value of this parameter in the determination road model for different parameters between the similar road model and the determination road model. Weighting is set for each edge component so that the weighting increases as these differences decrease.

  According to such a lane boundary detection device 10, it is possible to set a large weight to the edge component constituting the similar road model close to the determination road model, so that the boundary of the travel region close to the determination road model is It can be easily detected.

[Other Embodiments]
Embodiments of the present invention are not limited to the above-described embodiments, and can take various forms as long as they belong to the technical scope of the present invention.

For example, in the image processing of the present embodiment, the boundary of the travel region is detected using a known voting process such as a Hough transform, but a known process such as a least square method may be employed.
Furthermore, a process of converting the image coordinate system to a plane (road surface) coordinate system may be performed between the process of S330 and the process of S350 (S340). In this way, it is possible to perform processing such as calculating the amount of movement of the vehicle in the planar coordinate system, and to facilitate processing. In addition, processing by another application (for example, an application that determines whether or not to collide) can be facilitated. When the coordinate system is converted as described above, a road model or the like may be generated by converting a value in the plane coordinate system into an image coordinate system as necessary.

  Further, in the above-described embodiment, the processing in the case where the shape of the boundary of the travel region detected before (the shape of the similar road model) is a curve has been described. For example, the processing previously detected between S520 and S530 is performed. It is determined whether the shape of the boundary of the travel region thus obtained is a straight line or a curve, and if it is a straight line, in the process of S540, a point at infinity in the captured image (boundary of the travel region previously detected) The angle formed by the edge direction with respect to the direction of the virtual reference straight line representing the straight line connecting the position of the infinity point) and the position of the edge component may be calculated. In this way, the virtual reference straight line can be set with a simpler configuration.

  In the above embodiment, the similar road model is generated. However, the virtual reference straight line may be obtained only from the determination road model without generating the similar road model. In this case, in the process of S540, for example, a tangential direction may be obtained at a portion of the determination road model located at the same distance from the position of the edge component, and weighting may be performed according to the angle formed by this direction and the edge direction.

  DESCRIPTION OF SYMBOLS 1 ... Boundary detection system, 10 ... Lane boundary detection apparatus, 11 ... CCD camera, 12 ... ADC, 13 ... Image memory, 14 ... CPU, 15 ... ROM, 16 ... RAM, 20 ... Vehicle control ECU, 17 ... Communication IC.

Claims (6)

A boundary detection device that is mounted on a vehicle and detects a boundary of a travel region in which the host vehicle travels,
For each of a large number of pixels constituting a captured image obtained by imaging the periphery of the host vehicle, the magnitude of the luminance change between pixels adjacent in the vertical direction of the captured image and the magnitude of the luminance change between pixels adjacent in the horizontal direction are calculated. Edge direction calculating means for calculating an edge direction representing the direction in which the luminance changes the most by detecting and comparing the magnitude of the luminance change in each direction;
For each of the plurality of pixels, an edge component extraction unit that extracts an edge component representing a region where the luminance change between adjacent pixels is equal to or greater than a luminance reference value;
For each edge component, an angle calculation means for calculating an angle formed by the edge direction with respect to a direction of a virtual reference straight line representing a straight line connecting the position of the infinity point in the captured image and the position of the edge component;
First weight setting means for setting a weight for each edge component so that the weight becomes larger as the angle formed becomes closer to a right angle;
By performing image processing in consideration of the weighting set for each edge component, end detection means for detecting the end of the boundary of the travel region;
A boundary detection apparatus comprising: The boundary detection apparatus according to claim 1,
The boundary detection device is set to detect the boundary of the traveling region repeatedly,
Each time a boundary of the traveling area is detected, the position of a new infinity point in the captured image is detected according to the position of the boundary of the traveling area, and the position of the previous infinity point is determined as the position of the new infinity point. Infinity point updating means for updating to,
The said angle calculating means calculates the said angle | corner using the position of a new infinity point, The boundary detection apparatus characterized by the above-mentioned. In the boundary detection apparatus according to claim 1 or 2,
A determination road model generation means for generating a determination road model indicating the shape of the boundary of the travel region expected in the captured image based on the shape of the boundary of the travel region detected previously;
For each position of the edge component, the angle calculation means sets a virtual reference line in a tangential direction at a part corresponding to the position of the edge component in the road model for determination, and A boundary detection device characterized by calculating an angle formed by an edge direction. In the boundary detection apparatus according to claim 3,
A similar road model generation unit that generates a similar road model that passes through the position of the edge component for each position of the edge component and that differs only in some of the parameters from the plurality of parameters that form the determination road model; With
The angle calculation means sets, for each position of the edge component, a virtual reference line in a tangential direction in a part corresponding to the position of the edge component in the similar road model, and the edge with respect to the direction of the virtual reference line A boundary detection device characterized by calculating an angle formed by a direction. The boundary detection apparatus according to claim 4,
For parameters that are different between the similar road model and the judgment road model, the value of the parameter in the similar road model is compared with the value of the parameter in the judgment road model, and weighting is performed as the difference decreases. A boundary detection apparatus comprising: a second weight setting unit configured to set a weight for each edge component so that the   The boundary detection program for functioning a computer as each means which comprises the boundary detection apparatus of any one of Claims 1-5.