JP2008215836A - Object detection apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To prevent the occurrence of incorrect detection of objects due to travel environments. <P>SOLUTION: An ECU 100 executes object detection processing in a vehicle 10. In the processing, an image sensor 300 images a region in front of the vehicle 10 to generate image data. A pattern detection part 130 detects edge patterns on the basis of images indicated by the image data. In the case that edge patterns judged as those of objects are present among detected edge patterns, a control part 110 basically extracts the patterns as obstructions. When it is determined that the vehicle 10 approaches a structure which can cause significant changes in brightness in an imaging region of the image sensor 300 such as a tunnel entrance and a tunnel exit, it is determined that pattern detection accuracy by the pattern detection part 130 is not sufficiently secured, and patterns corresponding to detected objects are not extracted as objects but ignored. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a technical field of an object detection device that is mounted on a vehicle and detects various objects such as obstacles and white lines.

  As this type of apparatus, an apparatus including an imaging unit such as a camera has been proposed (for example, see Patent Document 1). According to the image processing apparatus disclosed in Patent Document 1 (hereinafter referred to as “conventional technology”), the position of the tunnel is acquired by a position detection unit including a self-supporting navigation system and the amount of incident light of the camera is adjusted. Therefore, it is said that a good image can be obtained even if the surroundings suddenly becomes dark.

  In addition, the technique which acquires the position of a tunnel with a navigation apparatus and determines a light lighting timing is also proposed (for example, refer patent document 2).

  In addition, a technique for detecting a tunnel by image processing and changing an exposure amount for obtaining a white line has been proposed (see, for example, Patent Document 3).

  Further, in the technical field of accident storage devices, it has been proposed to record video with different exposure times in order to be able to cope with a bright spot and a dark spot simultaneously (for example, Patent Document 4).

  Furthermore, in order to cope with a sudden change in brightness, a technique for adjusting the sensitivity at high speed by calculating and varying the sensitivity of the image element from a digital value subjected to A / D conversion has been proposed (for example, Patent Document 5).

Japanese Patent Laid-Open No. 11-42972 JP 2002-310671 A JP-A-11-205663 JP-A-9-205646 JP-A-7-120257

  When the amount of incident light is adjusted, it is possible to obtain a good image according to the change in ambient brightness, but it is possible to obtain a good image with the brightness with which the amount of incident light is suitable. For example, if the incident light quantity is adjusted to the dark part, the image quality of the bright part is deteriorated, and if the incident light quantity is adjusted to the bright part, the image quality of the dark part is obviously deteriorated. Therefore, when the conventional technique is applied to the detection of an object, the detection accuracy of the object may be extremely lowered at a place where the brightness changes drastically. That is, the conventional technique has a technical problem that an erroneous detection of an object may occur depending on a traveling environment.

  This invention is made | formed in view of the problem mentioned above, and makes it a subject to provide the target object detection apparatus which can prevent the misdetection of a target object.

  In order to solve the above-described problem, an object detection apparatus according to the present invention is an object detection apparatus that detects a predetermined type of object in a vehicle including an imaging unit capable of imaging a front detection range, Detection means capable of detecting a pattern corresponding to the object in the image of the detection range obtained via the imaging means, and presence for specifying the presence state of the object in the front based on the detection result of the pattern Identifies the relative positional relationship between the state specifying means, the vehicle, and a predetermined type of structure that is present in the travel route of the vehicle and set in advance to increase the degree of change in brightness in the detection range. Based on the result of the determination, a positional relationship specifying unit that determines whether the vehicle is approaching the structure based on the specified positional relationship, Determining means for determining a reflection mode when the detection result of the pattern is reflected in the specification of the presence state, and the presence state specifying unit is further determined when the reflection mode is determined. The presence state is specified based on a reflection mode.

  The vehicle according to the present invention is provided with imaging means that can take various aspects such as a monaural or stereo image sensor, a monaural or stereo CCD camera, or a video camera, and a predetermined detection range set in front of the vehicle. Can be imaged. Note that “imaging the detection range” is a concept including an image in which a space corresponding to the detection range can take various modes depending on the physical, mechanical, mechanical, or electrical configuration of the imaging means. For example, this means that the image is recorded in the form of physical or electrical data, regardless of whether it is temporary or permanent, and whether it is volatile or non-volatile.

  The “object” according to the present invention refers to various DSS (Driving Support System) that can be linked to the object detection apparatus according to the present invention, such as PCS (Pre Crush Safety system). It is a concept that encompasses objects that are related to the operational availability of safety systems or driving load reduction systems such as ACC (Adaptive Cruise Control), LKA (Lane Keeping Assist), or IPA (Intelligent Parking Assist). Obstacles or candidates for obstacles, including white objects (a color different from white) And other lane marks such as lane blocks.

  The “obstacle” typically refers to an object that is present on the traveling route of the own vehicle and that should be avoided by the subject vehicle. The object detection device defines the operation states of the various DSSs described above. In view of being able to do so, as a preferable aspect, a situation where the driver can be determined not to perform the avoidance operation, a situation where the driver is determined not to perform the avoidance operation, a situation where the driver's avoidance operation is not in time, Alternatively, it refers to the object to be avoided in a situation where the driver is not aware of the object to be avoided. In other words, when the driver is in a physical or mental situation where an avoidance operation can be performed, and when there is sufficient time and space to allow the driver to perform an avoidance operation, the driver's avoidance operation can be determined to be in time In other words, in a normal state or a period equivalent thereto, these objects to be avoided are objects that can become obstacles, that is, “obstacle candidates”, as a preferred embodiment. Therefore, the obstacle candidate and the obstacle are concepts that can point to the same object from the viewpoint of the physical properties of the object. The object according to the present invention can naturally include such an obstacle and an obstacle candidate.

  According to the object detection device of the present invention, for example, the detection unit such as an image processing device is provided, and the detection described above obtained through the imaging unit through various image processing such as edge detection processing and edge pattern detection processing, for example. It is configured to be able to detect a pattern corresponding to an object in a range image. Here, the “pattern corresponding to the object” refers to, for example, an edge pattern that can satisfy the criteria set as representing the above-described various objects.

  As a result of such pattern detection, for example, when the object is a lane mark, the concept includes, for example, the thickness and length of a white line, or a line type such as a solid line, a broken line, or a double line. Is detected in a form corresponding to the position of the vehicle (for example, a direction based on the vehicle, a distance from the vehicle estimated from an installation mode of the imaging unit, a setting mode of the detection range, and the like). For example, it is used as pattern data unique to the detected pattern, such as the number of edges, pattern shape or coordinate position. Alternatively, it is detected as the presence / absence of a pattern that can be regarded as a target object without including such pattern-specific information. The detection of such a pattern is performed, for example, in advance, or variably set by, for example, an operation of a driver or the like, for example, continuously or discretely on the time axis.

  According to the object detection device of the present invention, during its operation, for example, various processing units such as an ECU (Electronic Control Unit), various controllers or various computer systems such as a microcomputer device can be used. The specifying unit specifies the presence state of the object in front of the vehicle based on the detection result of the pattern as described above obtained through the detecting unit. Here, the “existing state” is a concept that encompasses a binary state whether or not it exists and a state that can be expressed numerically, such as the probability that the object exists or can be considered to exist. .

  The presence state specifying means specifies, for example, that each detected pattern exists as an object when the pattern corresponding to the object is detected by the detecting means during the operation. When the pattern corresponding to the object is not detected, or when the pattern detected at the previous detection timing is not detected, it is specified that the object does not exist (that is, the binary existence state is specified). Alternatively, the presence state specifying means gradually increases the existence probability of each detected pattern each time it is detected, for example, when a pattern corresponding to the object is detected by the detection means during the operation. Thus, the existence state is specified stepwise or continuously through a numerical index. Alternatively, similarly, when each detected pattern is not detected, for example, each of the existence probabilities is numerically reduced by gradually decreasing each time each pattern is not detected (for example, at each pattern detection timing). The presence state is specified stepwise or continuously through the index.

  As described above, in the presence state specifying unit according to the present invention, the pattern detection result is reflected in the presence state specification, and as a whole, the presence of an object corresponding to the pattern is detected as the pattern is detected. However, as to how the pattern detection result is reflected in the identification of the existence state, at least as long as the existence state of the object can be specified to the extent that no practical problems occur, for example, Thus, it may be free, including handling the detected pattern itself as an existing object.

  On the other hand, according to the object detection device of the present invention, during operation, for example, the vehicle and the vehicle by the positional relationship specifying means that can take the form of various processing units such as an ECU, various controllers or various computer systems such as a microcomputer device, A relative positional relationship with a predetermined type of structure that is present in the travel route of the vehicle and set in advance to increase the degree of change in brightness in the detection range described above is specified.

  Here, the “structure” according to the present invention is present in the travel route of the vehicle as a concept including a form on the travel route of the vehicle or in the vicinity of the travel route, and reaches or arrives when the vehicle approaches. It is a concept that encompasses objects that are set as being capable of causing a large change in brightness when viewed relatively in the detection range, and as a preferable aspect, for example, an entrance / exit part of a tunnel, an entrance / exit under a guard of a railway Or the edge part (namely, the part which prescribes | regulates the start point and end point of these installation sections) etc., such as a sound insulation wall or a rock fall prevention board. In addition, “relatively large brightness change” described here means, more specifically, that the image quality of the image obtained through the imaging unit is related to the pattern detection accuracy by the detection unit. It means that there is a possibility of bringing about a change in brightness that can be deteriorated to such an extent that it can be affected.

  Further, the “relative positional relationship” according to the present invention is a qualitative positional relationship such as whether the distance between the vehicle and the structure is large or small, for example, the distance between the vehicle and the structure is higher than a certain reference value. It is a concept that encompasses a binary positional relationship such as whether it is large or small and a quantitative positional relationship such as how many meters (or kilometers) the distance between the vehicle and the structure is.

  The positional relationship specifying means may be, for example, the absolute position (latitude, longitude and altitude) of the current vehicle, or own vehicle position information indicating a position on a map stored as data in an appropriate storage means, a general road, an automobile exclusive road Or information on the type of expressway, information on speed limit, information on legal regulations such as one-way traffic, information on temporary regulations such as construction, lane information, and tunnels, bridges, sound insulation boards, railroad crossings Provide navigation information that can appropriately include various infrastructure information such as the position of guards, rockfall prevention plates, etc., or destination information on places that can be destinations such as stations, hospitals, schools, shopping centers, stadiums or companies For example, an electric signal and a physical or physical navigation device that can be mounted on a vehicle using a positioning system such as a GPS (Global Positioning System) that can be configured. Obtains information to be used for specifying such a relative positional relationship as electrical data, and specifies the relative positional relationship. Alternatively, the positional relationship specifying means acquires a detection signal transmitted from a transmitting device or the like installed as an infrastructure facility at a tunnel entrance and the like, and specifies the relative positional relationship.

  Here, in particular, when the vehicle travels near or around the structure described above, for example, when approaching the structure, reaching the structure, or passing through the structure, the detection range is relatively For example, an excessively bright part and an excessively dark part, for example, a situation where a pattern that has been detected until then is suddenly not detected (disappearance of a pattern) or a pattern that has not been detected is detected suddenly. Things can happen. Of course, these are not necessarily all errors caused by such a change in brightness, but in relative terms, such a change in brightness within the detection range is too large, so that an object that does not exist Misdetection of a pattern is easy to occur as a concept that includes detection of a pattern as well as detection of an object that should be detected as a pattern. At this time, the presence state specifying means specifies the presence state in a manner that reflects the detection result of the pattern as already described. Therefore, in this case, the detection accuracy of the target object is likely to deteriorate.

  Therefore, according to the object detection device according to the present invention, the relative relationship specified by the positional relationship specifying means by the determining means that can take the form of various processing units such as an ECU, various controllers or various computer systems such as a microcomputer device, for example. Whether or not the vehicle is approaching the structure is determined based on the positional relationship. Further, when the pattern detection result is reflected in the identification of the presence state based on the result of the determination by a determination unit that can take the form of various processing units such as an ECU, various controllers or various computer systems such as a microcomputer device, etc. The reflection mode is determined. When the reflection mode is determined, the presence state specifying unit specifies the presence state based on the reflection mode.

  Here, the criterion for determining whether or not the vehicle is approaching the structure is that the detection accuracy of the object is reduced due to the decrease in the pattern detection accuracy due to the approach to the structure. It is not limited at all as long as it is determined to be able to suppress the decrease in the detection accuracy to some extent compared with the case where no countermeasures are taken, for example, experimentally, empirically, theoretically or simulation in advance. Based on the above, it is determined that the vehicle is approaching the structure when the vehicle approaches the structure to such an extent that a decrease in pattern detection accuracy can be practically manifested due to a change in brightness within the detection range. It may be determined to be made. Alternatively, the criterion is that the vehicle is further away from the structure to such an extent that even if a change in brightness occurs in the detection range, a decrease in the detection accuracy of the object is suppressed to a range that does not manifest at least in practice. May be determined so that it is determined that the object is not approaching. As a preferred embodiment, the determination criterion is set as a fixed or variable reference value that is determined for each type of the structure or that is uniformly determined regardless of the structure. That is, in this case, the presence or absence of approach is determined according to the magnitude relationship with the reference value.

  On the other hand, the “reflection mode” determined by the determination unit is a stepwise or continuous process such as a binary mode for determining whether or not the pattern detection result is reflected in the presence state specification, and the degree of reflection in the presence state specification. Specific aspects may be taken. In other words, the determination unit weights the pattern detection result according to the determination result related to the determination unit. The sensitivity of the presence state to the pattern detection result changes in a binary manner, stepwise, or continuously by determining the reflection mode and specifying the presence state according to the determination.

  Here, such a reflection mode need not always be determined in specifying the presence state, and may be determined only when the vehicle is approaching the structure, for example. In other words, except for limited conditions mainly including approach to the structure, the pattern detection result is basically a process for identifying the presence state that can be normally performed by the presence state identifying means (for example, every time a pattern is detected). The reflection mode may not be intentionally set so that it is provided as usual for a process such as specifying that it exists as an object or increasing the probability of being an object every time a pattern is detected. . Of course, substantially the same effect can be obtained even if the reflection mode is determined such that the detection result of the pattern is used to specify the presence state with a degree of reflection that is 100% or high enough to be regarded as it. In such a case, in view of the fact that the detection accuracy of the pattern may be reduced with the approach to the structure, the determination means determines that the subject vehicle is approaching the structure. The reflection mode may be determined, for example, such that the specified existence state is relatively difficult to change (that is, less likely to be affected by the pattern detection result).

  Thus, according to the object detection device according to the present invention, for example, when the vehicle approaches a structure such as a tunnel doorway, the reflection mode of the pattern detection result when specifying the existence state of the object is provided. Be changed. Thus, for example, due to the change in brightness caused by the dark part inside the tunnel near the tunnel entrance and the bright part around the tunnel, or for example near the tunnel exit, the bright part near the exit and the dark part inside the tunnel. The occurrence of false detection of an object due to a decrease in pattern detection accuracy due to a change in brightness caused by.

  In one aspect of the object detection apparatus according to the present invention, the structure includes at least one of an entrance portion and an exit portion in the tunnel.

  Since the tunnel is structurally dark inside and the surroundings are likely to be bright, the portion corresponding to the entrance / exit of the tunnel is likely to have a noticeable deviation in brightness within the detection range, and erroneous detection of the object is likely to occur. Therefore, when either one or preferably both of the entrances and exits of the tunnel are adopted as the structure as described above, the object detection device according to the present invention can act remarkably effectively.

  In another aspect of the object detection apparatus according to the present invention, the vehicle further includes position detection means for detecting absolute positions of the vehicle and the structure, and the positional relationship specifying means is configured to detect the detected absolute position. The positional relationship is specified based on

  According to this aspect, since the vehicle is provided with the position detection unit that detects the absolute position of the vehicle and the structure, such as the car navigation device described above, the relative positional relationship between the vehicle and the structure is determined by the positional relationship specifying unit. Can be easily specified with high accuracy.

  In another aspect of the object detection apparatus according to the present invention, the presence state specifying unit specifies the presence probability of the target as the presence state.

  According to this aspect, since the existence probability as the probability that the object exists or can be considered to exist is specified as the existence state of the object, the object exists or does not exist, etc. It is possible to grasp the presence of the object more flexibly and accurately than when the presence state is specified in a binary manner. For example, when the certainty of the presence of an object can be expressed numerically in this way, it is possible to reflect the detection result relating to pattern detection that can be performed at a constant or indefinite period in the numerical value. That is, the presence state can be specified while reflecting the detection result of the past pattern within a finite time range. Therefore, the above-described various safety systems that can be linked to the object detection device can be operated more efficiently and effectively.

  In another aspect of the object detection apparatus according to the present invention, the positional relationship specifying means specifies a distance between the vehicle and the structure as the positional relationship, and the determining means has the specified distance as a predetermined distance. When the value is less than the value, the determination that the structure is approaching is performed.

  According to this aspect, the positional relationship between the vehicle and the structure is specified as the distance between them by the positional relationship specifying means. The distance is an index that can easily and accurately determine whether or not the vehicle is approaching the structure, and the accuracy of determination related to the determination unit is easily ensured. At this time, the predetermined value used for the determination may be a fixed value or a variable value. When the predetermined value is a variable value, for example, the type of the structure or the vehicle It may be a variable value according to the vehicle speed (including a case where the arrival time at the structure is a fixed value).

  In another aspect of the object detection device according to the present invention, the determination means determines whether the detection result of the pattern is the presence state when the determination is made that the structure is approaching. The reflection mode is determined so as not to be reflected.

  According to this aspect, when the determination means that the vehicle is approaching the structure is made, the determining means uses the reflection aspect described above so that the pattern detection result is less likely to be reflected in the presence state specification. Therefore, it is possible to effectively prevent specification of an erroneous presence state (that is, erroneous detection of an object) based on a detection result of a pattern whose detection accuracy is difficult to be ensured.

  It should be noted that the aspect of making it difficult to reflect the detection result of the pattern in the identification of the existence state is not limited as long as the influence of the detection result on the identification of the existing state can be reduced as a whole. The reflection mode may be determined such that no reference is made. In this case, the presence state specifying means maintains the presence state of the target object as the specification result before the vehicle approaches the structure, that is, based on the detection result of the pattern whose detection accuracy is ensured. Good. Alternatively, the presence state specifying means may detect an erroneous pattern based on the presence state of the target object for which accuracy is ensured, for example, experimentally, empirically, theoretically, or based on simulation. Rather than specifying the presence state based on the above, by performing numerical operations or logical operations according to algorithms, calculation formulas or theoretical formulas that are set so that the presence state can be accurately shifted, The presence state may be estimated.

  Alternatively, the determining means specifies the presence state of the pattern detection result by relatively reducing the increase amount or the increase rate or the decrease amount or the decrease rate when increasing or decreasing the existence probability of the target object. It may be difficult to be reflected in.

  In one aspect of the object detection apparatus according to the present invention in which the reflection mode is determined so that the pattern detection result is difficult to be reflected, the determination unit determines that the object is approaching the structure. In this case, the reflection mode is determined so that the detection result of the pattern is not reflected in the specification of the existence state.

  The presence of the structure can significantly affect the pattern detection result, but is often unrelated to the actual state of the object. Therefore, if a pattern is detected as a conventional state, there is a high possibility that an object corresponding to the pattern exists even after approaching the structure. Similarly, if a pattern is not detected as a previous state, there is a high possibility that an object corresponding to the pattern does not exist even after approaching the structure. On the other hand, after approaching the structure, the pattern detection environment is worse than before approaching, so the newly detected pattern after approaching the structure is a pattern corresponding to an object that does not originally exist, that is, There is a high possibility that the pattern will lead to erroneous detection of the object. Similarly, a pattern that is no longer detected after approaching the structure is likely to be a pattern that could not be detected despite the presence of the target object.

  Therefore, in the state where the vehicle is approaching the structure in this way, the pattern detection result is not reflected in the identification of the existence state of the object, so that the existence state of the object already obtained can be trusted. It is possible to prevent the deterioration of the quality, and it is possible to prevent the quality deterioration of the entire object detection apparatus.

  Note that in this aspect, the determination means determines that the detection result of the pattern is in the presence state when the same pattern is detected before and after the determination that the structure is approached. The reflection mode may be determined so as to be reflected specifically.

  While approaching a structure, the pattern detection accuracy is basically difficult to ensure, that is, it is difficult to guarantee that the pattern detection accuracy is sufficiently secured for reasonable objective reasons. However, if the same pattern (or that can be considered or judged to be the same) is detected as a conventional state, that is, before approaching the structure, the presence of the object itself is such a structure. In view of the fact that it is difficult to be influenced by the presence or absence of the object, it can be determined that the detected pattern accurately represents the existing object. Therefore, in such a case, by accurately reflecting the detection result of the pattern in the identification of the presence state, it is possible to accurately identify the presence state of the target object based on the detected pattern while preventing erroneous detection of the target object. It is possible to continue as much as possible, which is beneficial in practice.

  In another aspect of the object detection apparatus according to the present invention in which the reflection mode is determined so that the pattern detection result is difficult to be reflected, the presence state specifying means determines that the object is approaching the structure. When the determination is made, the presence state is identified based on the presence state identified before the determination that the structure is approaching.

  According to this aspect, in a situation where at least a part of the pattern detection result is not reflected in the identification of the presence state, the presence state of the object is based on the presence state specified in the past (that is, before approaching the structure). Is identified. That is, in this case, the presence state specifying means substantially estimates the presence state of the object. In this case, the estimation of the presence state of the object is, for example, experimentally conducted in advance so that the identification accuracy can be ensured more than when the erroneous presence state is specified based on at least the erroneously detected pattern. In particular, it may be made through an algorithm that can be set theoretically or based on a simulation, a numerical operation or a logical operation based on a calculation equation, a logical equation, or the like.

  According to this aspect, even if the structure temporarily falls short of the accuracy of detection of the pattern corresponding to the object, if the object actually exists, the presence state is specified by the presence state specifying means. Therefore, the presence of the vehicle is not erased, and the safety of the vehicle is ensured.

  In another aspect of the target object detection apparatus according to the present invention, the reflection mode is determined so that the pattern detection result is difficult to be reflected. The presence state specifying unit specifies the presence probability of the target object as the presence state. In addition, each time the pattern is detected, the existence probability is increased by a first correction amount, and when the determination means determines that the structure is approaching, the first means The correction amount is corrected to the decreasing side.

  According to this aspect, the presence state specifying means is configured to be able to specify the presence state of the object as the presence probability as already described. At this time, the presence state specifying means increases the presence probability by the first correction amount every time a pattern corresponding to the object is detected. That is, according to this aspect, every time a pattern corresponding to the object is detected, the existence probability of the object gradually increases (of course, since it is a probability, the existence probability is preferably set to 100% as the upper limit. To increase).

  On the other hand, after approaching the structure, that is, when it can be determined that the detection accuracy of the pattern is considerably lowered, the determining means corrects the first correction amount to the decreasing side. Therefore, in the state of approaching the structure, even if a pattern is detected, the existence probability is unlikely to increase, and the sensitivity of the existence probability to the pattern detection result decreases as a whole. That is, according to this aspect, the reflection aspect is determined such that the first correction amount is corrected to the decreasing side, so that the pattern detection result is not easily reflected in the specification of the presence state of the object.

  According to this aspect, before the approach to the structure (that is, the situation in which the pattern detection accuracy is ensured, for example, after the approach to the structure is finished (for example, during passage through the tunnel) or during normal travel) In other words, the pattern detection result is reflected in specifying the existence state with a certain degree of reliability even if reliability is not set. Therefore, there is a practically high effect of positively specifying the presence state of the target object without significantly reducing the accuracy of specifying the target state. It should be noted that the period of time when it is determined that the object is approaching is often not so long in practice, and the presence state of the object is determined based on the detection result of the pattern whose detection accuracy is difficult to be ensured. Even if it is specified, it is difficult to cause problems in practice. Moreover, even if the reduction amount of the first correction amount is a fixed value, it can be set by the user, or a variable value that can be appropriately given by the determination means in consideration of the driving conditions of the vehicle each time. If it is variable, it is possible to prevent the existence probability from changing greatly compared to before the approach to the structure in practice by setting the reduction amount relatively large.

  In another aspect of the target object detection apparatus according to the present invention, the reflection mode is determined so that the pattern detection result is difficult to be reflected. The presence state specifying unit specifies the presence probability of the target object as the presence state. In addition, every time the pattern is not detected, the existence probability is decreased by a second correction amount, and when the determination means determines that the structure is approaching, the second means Correct the correction amount to the decreasing side.

  According to this aspect, the presence state specifying means is configured to be able to specify the presence state of the object as the presence probability as already described. At this time, the presence state specifying means decreases the presence probability by the second correction amount every time a pattern corresponding to the object is not detected. That is, according to this aspect, every time a pattern corresponding to an object is not detected, the existence probability of the object gradually decreases (of course, since it is a probability, the existence probability is preferably reduced with 0% as the lower limit). To do).

  On the other hand, after approaching the structure, that is, when it can be determined that the detection accuracy of the pattern is considerably lowered, the determination unit corrects the second correction amount to the decrease side. Therefore, in the state of approaching the structure, even if the pattern is not detected, the existence probability is difficult to decrease, and the sensitivity of the existence probability to the pattern detection result is reduced as a whole. That is, according to this aspect, the reflection aspect is determined such that the second correction amount is corrected to the decreasing side, so that the pattern detection result is not easily reflected in the specification of the presence state of the object.

  According to this aspect, before the approach to the structure (that is, the situation in which the pattern detection accuracy is ensured, for example, after the approach to the structure is finished (for example, during passage through the tunnel) or during normal travel) In other words, the pattern detection result is reflected in specifying the existence state with a certain degree of reliability even if reliability is not set. Therefore, there is a practically high effect of positively specifying the presence state of the target object without significantly reducing the accuracy of specifying the target state. It should be noted that the period of time when it is determined that the object is approaching is often not so long in practice, and the presence state of the object is determined based on the detection result of the pattern whose detection accuracy is difficult to be ensured. Even if it is specified, it is difficult to cause problems in practice. Moreover, even if the reduction amount of the second correction amount is a fixed value, it can be set by the user, or a variable value that can be appropriately given by the determining means in consideration of the driving conditions of the vehicle each time. If it is variable, it is possible to prevent the existence probability from changing greatly compared to before the approach to the structure in practice by setting the reduction amount relatively large.

  Such an operation and other advantages of the present invention will become apparent from the embodiments described below.

<Embodiment of the Invention>
Hereinafter, various preferred embodiments of the present invention will be described with reference to the drawings as appropriate.

<First Embodiment>
<Configuration of Embodiment>
First, the configuration of the vehicle 10 according to the first embodiment of the present invention will be described with reference to FIG. FIG. 1 is a block diagram conceptually showing a configuration of a part related to the object detection device according to the present invention in the vehicle 10.

  In FIG. 1, a vehicle 10 includes an ECU 100, a navigation device 200, and an image sensor 300.

  The ECU 100 is configured to control the operation of each component of the vehicle 10 including an engine (not shown) and to execute an object detection process to be described later according to a control program. By executing the object detection process, the ECU 100 functions as an example of the “object detection device” according to the present invention.

  The ECU 100 includes a control unit 110, a position information storage unit 120, a pattern detection unit 130, and an object information storage unit 140.

  The control unit 110 includes a CPU (Central Processing Unit) and a ROM (Read Only Memory) (not shown), and is a main control unit configured to be able to control the entire operation of the ECU 100. It is configured to be able to function as an example of each of “existing state specifying means”, “positional relationship specifying means”, “discriminating means”, and “deciding means”.

  The position information storage unit 120 is a part corresponding to one storage area in a volatile storage device such as a RAM (Random Access Memory) or a buffer memory. The position information storage unit 120 is configured to store position information data supplied from the navigation device 200.

  The pattern detection unit 130 is an image processing unit configured to be able to detect an edge pattern corresponding to an object from an image captured by the image sensor 300, and is an example of the “detection unit” according to the present invention. is there.

  Note that the object in the present embodiment adopts various aspects depending on the type of DSS mounted on the vehicle 10, and when the detection result of the object is used in a safety system such as a PCS, Is preferably an obstacle (including obstacle candidates) such as a forward vehicle or a falling object on the road, and is suitable as an object when the detection result is used in a driving load reduction system such as ACC, LKA, or IPA. For example, various lane marks such as white lines are adopted. However, the effects according to the present embodiment are similarly achieved regardless of the type of the object. In the following description, unless otherwise specified, the object and the object are simply distinguished. I will express it.

  Similar to the position information storage unit 120, the object information storage unit 140 is a part corresponding to one storage area in a volatile storage device such as a RAM or a buffer memory. The object information storage unit 140 is configured to store information related to the object detected in the object detection process (to be described later) via the control unit 110 (that is, the size and position of the object).

  The navigation device 200 includes various positioning systems such as GPS on a map screen displayed on various display means such as a liquid crystal display panel based on map data stored in an appropriate storage means such as an HDD (Hard Disk Drive). 3 is an example of a “position detecting unit” according to the present invention configured to be able to display position information based on an absolute position obtained through the position information. The navigation device 200 is electrically connected to the ECU 100, and data relating to the position of the vehicle 10, more specifically, data relating to the absolute position of the vehicle 10, data relating to the absolute position of structures existing around the vehicle 10, and the like. Regardless of whether or not the map screen is displayed on the display means, the ECU 100 is constantly or as long as the positioning system is in operation or at a constant or indefinite period.

  The image sensor 300 includes a monocular imaging unit that is installed at a predetermined position in front of the vehicle 10 so as to image a front region of the vehicle 10 (that is, an example of the “detection range” according to the present invention) and an image. 3 is an example of an “imaging unit” according to the present invention, including an image generation unit that generates image data relating to an image in the detected range. The image data generated in the image sensor 300 is sent to the pattern detection unit 130 at a detection cycle controlled by the control unit 110.

<Operation of Embodiment>
Next, with reference to FIG. 2, the details of the object detection processing executed by the ECU 100 will be described as the operation of the present embodiment. FIG. 2 is a flowchart of the object detection process.

  In FIG. 2, the control unit 110 controls the image sensor 300 so that the front area of the vehicle 10 is imaged (step A10). At this time, an image signal corresponding to the detection range imaged by the image sensor 300 is converted into image data inside the image sensor 300 and sent to the pattern detection unit 130. Note that such imaging of the detection range is repeatedly executed at a detection cycle preset by the control unit 110.

  When the image data is transmitted, the pattern detection unit 130 detects an edge pattern in the transmitted image data (step A11). In the process related to step A11, the pattern detection unit 130 performs each process such as edge point extraction, coordinate conversion, and pattern detection to detect an edge pattern.

  Edge point extraction is a process of extracting edge points from image data using, for example, a first-order differential filter. The coordinate conversion process is a process of associating the extracted edge point with the coordinate position in the image represented by the image data. When the object is various lane marks such as white lines, the coordinate conversion process is preferably executed so that the image represented by the image data is an image looking down on the travel route of the vehicle 10 from directly above. The Pattern detection is performed, for example, on whether the extracted edge point is a down edge (for example, an edge where the brightness of the image decreases by a predetermined value or more) or an up edge (for example, an edge where the brightness of the image increases by a predetermined value or more). This is a process of executing discrimination for each extracted edge point and detecting the extracted edge point as an edge pattern. When the object is a white line or the like, a straight line approximation of an edge point by Hough transformation or the like may be performed at this time. By executing each process in this way, an edge pattern is appropriately detected for one image.

  In the process related to step A11, the pattern detection unit 130 generates pattern data including information (number of edge points, pattern shape, and coordinate position) unique to the detected pattern, and supplies the pattern data to the control unit 110. Note that the processing related to edge detection described here is an example, and various known methods can be used for edge detection.

  When the pattern detection ends, the control unit 110 determines whether or not the detected pattern is a pattern corresponding to the object (step A12). For example, for each detected edge pattern, the control unit 110 compares, for example, a threshold value provided for each target object with the number of edges, and when there is an edge pattern having an edge number equal to or greater than the threshold value, the edge The pattern is determined as an edge pattern corresponding to the object (that is, an example of “a pattern corresponding to the object” according to the present invention). Note that the pattern detection unit 130 may determine whether the pattern corresponds to the object. In this case, the above-described pattern data may be generated and supplied to the control unit 110 only for the pattern that has been determined to correspond to the object, or may be temporarily stored in the RAM or buffer memory. Good.

  When there is no pattern corresponding to the object (step A12: NO), the control unit 110 shifts the process to step A16 and executes an interpolation process described later. On the other hand, if there is a pattern corresponding to the object (step A12: YES), the control unit 110 sets the relative distance L of the tunnel entrance to the vehicle 10 (that is, an example of the “structure” according to the present invention). (Step A13).

  When executing the object detection process, the control unit 110 detects the travel position of the vehicle 10 from the navigation device 200 and the position of a tunnel (that is, an example of the “structure” according to the present invention) that exists in the travel route of the vehicle 10. Exactly, the position of the entrance / exit of the tunnel) is acquired as position data. The acquired position data is stored in the position information storage unit 120. In the processing according to step A13, the control unit 110 calculates and sets the relative distance L based on the stored position data.

  In the process according to step A13, it is only necessary to determine whether or not the vehicle 10 is approaching the tunnel entrance (or exit), so that it is unnecessarily far away (that is, does not substantially affect the pattern detection result). It is not necessary to consider the existence of tunnel entrances (distant). Therefore, the position information storage unit 120 only has a tunnel that is a travel route of the vehicle 10 and is within a predetermined distance (at least a distance that does not affect the pattern detection accuracy). Position data is supplied. At this time, if position data regarding the tunnel entrance is not stored in the position information storage unit 140, the control unit 110 sets the relative distance L to the maximum value Lmax. Of course, as long as the tunnel exists in the travel route of the vehicle, the position data regarding the tunnel may be supplied without providing the distance limitation.

  In this embodiment, an example of the “structure” according to the present invention is not limited to a tunnel entrance / exit, but under a railway guard (preferably an entrance / exit under the guard), a sound insulation wall, a protective wall, or a rock fall prevention plate. (Preferably, these installation start points and installation end points) can be similarly applied. In this case, the control unit 110 may acquire position data regarding such multiple types of structures from the navigation device 200. The relative distance L described above may be set individually for each of these structures, or may be set only for the structure closest to the vehicle 10. In addition, such setting of the relative distance L is an effective index when these various structures exist on the travel route of the vehicle 10 (that is, in front of the vehicle 10), and the vehicle 10 passes through these structures. At this point, more preferably, these structures are excluded from the relative distance setting targets when there is an objective and reasonable reason for ensuring the pattern detection accuracy.

  When the relative distance L with respect to the tunnel entrance is set, the control unit 110 determines whether or not the set relative distance L is less than a predetermined threshold Lth (step A14). The threshold value Lth is an example of the “predetermined value” according to the present invention. The threshold value Lth has a significant change in brightness such that a decrease in pattern detection accuracy by the pattern detection unit 130 can be manifested in an imaging region of the image sensor 300 (that is, the above-described front region), such as a tunnel entrance or exit. For example, in this embodiment, the distance is set to a value of about several tens of meters to several hundreds of meters. Needless to say, the threshold value Lth is set to be shorter than the above-described “predetermined distance” in which the position data regarding the tunnel entrance is stored in the position information storage unit 120.

  When the relative distance L is greater than or equal to the threshold value Lth (step A14: NO), the control unit 110 practically has detection accuracy related to pattern detection performed by the pattern detection unit 130 on an image obtained via the image sensor 300. As an example that can be secured without any problem, it is an example of a pattern that is determined to be an object in step A12 (that is, an example of a “pattern corresponding to an object” according to the present invention, and an example of a “pattern detection result”. Is extracted as an object (step A17). Here, “extracting as an object” means that a pattern determined to be an object in the process according to step A12 is formally regarded as an object (that is, “existence state according to the present invention” “Identify” as an example) indicates that the above-described pattern data corresponding to the pattern is stored in the object information storage unit 140. That is, in other words, in the process according to step A17, as an example of the “reflection state” according to the present invention, a reflection state for reflecting the pattern detection result in the identification of the existing state is determined. In the present embodiment, as described above, a binary state of whether to reflect is adopted as the reflected state of the detected pattern, and the presence state is specified as a binary state of whether or not it exists.

  The pattern extracted as an object in this manner is an object of continuous monitoring, and is used for control of various safety systems that can be provided in the vehicle 10, for example. For example, if the object is an obstacle, the corresponding pattern is extracted as an obstacle candidate in the process according to step A17.

  On the other hand, when the relative distance L is less than the threshold value Lth (step A14: YES), the control unit 110 has detection accuracy related to pattern detection performed by the pattern detection unit 130 on an image obtained via the image sensor 300. Assuming that it cannot be sufficiently secured, the presence of the pattern determined to be an object in step A12 is ignored (step A15), and interpolation processing is executed (step A16). Here, “ignore” means that the pattern is not reflected in the identification of the presence state of the object in front of the vehicle 10, that is, at this stage, as an example of the “reflection state” according to the present invention. The reflection mode that the pattern detection result is not reflected in the presence state specification is determined. When the pattern corresponding to the detected object is ignored, the pattern data related to the pattern may be discarded or temporarily stored in another storage area.

  When executing the interpolation processing according to step A16, the control unit 110 determines whether or not there is pattern data regarding a pattern stored as an object in the object information storage unit 140. If the pattern data does not exist, it is assumed that the presence state that the object does not exist before the approach to the tunnel has been specified, and the control unit 110 indicates that the object does not exist. Continue to identify. That is, in this case, the control unit 110 performs substantially no processing.

  On the other hand, when there is pattern data related to the pattern extracted as the object in the object information storage unit 140, the control unit 110 displays the coordinate position of the pattern corresponding to each of the pattern data stored in the object information storage unit 140. Is corrected to the coordinate position where the pattern (that is, the object) is estimated to exist at the present time based on the position information of the vehicle 10 stored in the position information storage unit 120, the vehicle speed, the steering information, and the like of the vehicle 10. Then, the object whose existence is estimated is newly stored in the object information storage unit 140 as estimated data. That is, in this case, the presence state of the object is estimated.

  Thus, when the process according to step A16 or step A17 is executed, the process returns to step A10, and a series of processes is repeated. When the presence of the object is estimated, the estimated state of the object (including the coordinate position) tends to deviate from the actual situation as the number of estimations increases. Therefore, when the estimated number exceeds the predetermined number, the interpolation process according to step A16 is prohibited, that is, the update of the estimated data related to the estimated state of the target object is stopped, and the reliability regarding the target state of the target object is stopped. It may be possible to secure the sex.

  It should be noted that when the vehicle 10 enters the tunnel (that is, when the vehicle 10 is no longer in the state of approaching the tunnel entrance) in the repeated process of such processing, the situation must be considered as approaching the tunnel exit. If the relative distance to the tunnel exit is equal to or greater than Lth, or if the relative distance L is set to Lmax as described above, the determination processing related to step A14 is “NO”, and the object is detected. Such a process quickly returns to the normal process, and the pattern corresponding to the object is quickly extracted as the object. At this time, the coordinate position is interpolated by the interpolation processing in step A16, and the estimated data relating to the pattern whose existence has been estimated is discarded.

  As described above, according to the object detection process according to the present embodiment, when the vehicle 10 is approaching the entrance of the tunnel, it is difficult to ensure the pattern detection accuracy by the pattern detection unit 130. The pattern corresponding to the object detected by the pattern detection unit 130 is ignored without being reflected in the process related to specifying the presence or absence of the object. Therefore, erroneous detection of the object is prevented. In this case, the object extraction result before the approach to the entrance of the tunnel, that is, in the situation where the pattern detection accuracy is ensured (that is, the determination of “approaching to the structure” Since the presence state of the target object in the period in which the pattern detection result is not reflected is estimated based on the “presence state specified before being performed”, the target monitoring itself is affected by the travel route of the vehicle 10. It is continuous and useful in practice.

Second Embodiment
Next, with reference to FIG. 3, the object detection process which concerns on 2nd Embodiment of this invention is demonstrated. FIG. 3 is a flowchart of the object detection process according to the second embodiment of the present invention. In the figure, the same reference numerals are given to the same portions as those in FIG. 2, and the description thereof will be omitted as appropriate.

  In FIG. 3, when the relative distance L between the vehicle 10 and the tunnel entrance is less than the threshold value Lth (step A14), the control unit 110 determines that the pattern determined to be an object in the process according to step A12 is It is determined whether or not the pattern is the same as the pattern stored as the object in the object information storage unit 140 (step B10). At this time, the control unit 110 refers to the pattern data related to the pattern of the object stored in the object information storage unit 140, and the pattern is determined to be the object in the process according to step A12. The discrimination processing is executed by comparing the number of edges, pattern shape, coordinate position, and the like.

  As a result, when the same pattern is stored in the object information storage unit 140 as a pattern corresponding to the object (step B10: NO), the control unit 110 shifts the process to step A17, The pattern determined to be is extracted as an obstacle. That is, the presence state of the object is updated. On the other hand, when the pattern corresponding to the object detected this time does not match any pattern stored in the object information storage unit 140 (step B10: YES), the control unit 110 moves the process to step A15. Then, the pattern is ignored, and the interpolation process described above is executed to estimate the presence state of the object.

  Thus, according to the object detection process according to the second embodiment, the pattern detection accuracy by the pattern detection unit 130 in the image obtained via the image sensor 300 is not ensured by approaching the entrance of the tunnel. Even so, if the same pattern is extracted as an object as a conventional state, that is, before approaching such a tunnel entrance, the detected pattern is not ignored, that is, The detected pattern is used to identify the presence state of the object without changing the reflection state, or the reflection state is set to the normal reflection state, or the reflection state is not set specially. The

  Whether or not the vehicle 10 is approaching various structures such as a tunnel can significantly affect the pattern detection accuracy by the pattern detection unit 130, but is completely independent of the actual presence state of the target object. Since there are many cases, when it is possible to detect a pattern that has already been extracted as an object, there is no need to ignore the detection result of such a pattern. As described above, according to the second embodiment, erroneous detection is caused due to the approach to various structures such as a tunnel while continuing to specify the existence state of the object reflecting the pattern detection result as much as possible. It is possible to reliably prevent a situation in which the presence state of the object is specified based on the pattern that has been made.

<Third Embodiment>
In the first and second embodiments described above, the presence state of the object is specified as a binary state indicating whether or not the object exists, but the presence state of the object may take other forms. Here, with reference to FIG. 4, the object detection process which concerns on 3rd Embodiment of this invention is demonstrated. FIG. 4 is a flowchart of the object detection process according to the third embodiment. In the figure, the same reference numerals are assigned to the same portions as those in FIG. 2, and the description thereof is omitted as appropriate.

  In FIG. 4, the control unit 110 sets a relative distance L (step A13). Next, the control unit 110 sets the counter up amount Dup to Dup1 (step C10), and sets the counter down amount Ddn to Ddn1 (step C11). Here, the counter-up amount Dup and the counter-down amount Ddn are correction amounts for calculating the existence probability of the object, respectively, and are “first correction amount” and “second correction amount” according to the present invention, respectively. It is an example.

  Next, the controller 110 determines whether or not the relative distance L is less than the threshold value Lth (step A14), as in the above-described embodiments. If the relative distance L is less than the threshold value Lth (step A14) : YES), that is, in a situation where the pattern detection accuracy related to the pattern detection unit 130 is not ensured, the counter-up amount Dup is reset to Dup2 (Dup2 <Dup1) (step C12). Similarly, the counter-down amount DdnwoDdn2 (Ddn2 <Ddn1) is reset (step C13). On the other hand, when the relative distance L is greater than or equal to the threshold Lth (step A14: YES), the control unit 110 does not reset the counter Dup and the counter Ddn. In this way, at the stage where the counter up amount Dup and the counter down amount Ddn are set to Dup1 or Dup2 and Ddn1 or Ddn2, respectively, the control unit 110 executes the processing related to Step A10 to Step A12.

  As a result, when there is a pattern corresponding to the object among the detected edge patterns (step A12: YES), the control unit 110 determines the object existence probability R (i) (i represents the number of trials). (Natural number) is calculated according to the following equation (1) (step C14). When there is no pattern corresponding to the object among the detected edge patterns (step A12: NO), the control unit 110 calculates the object existence probability R (i) according to the following equation (2). Calculate (step C15). In these equations, R (i-1) is the existence probability of the object at the previous pattern detection timing.

R (i) = R (i-1) + Dup (1)
R (i) = R (i-1) -Ddn (2)
As described above, when it is determined that there is a pattern corresponding to the object at one pattern detection timing, the control unit 110 calculates the existence probability of the object corresponding to the pattern last time. The existing probability is updated by adding the counter-up amount Dup with an upper limit of 100% to increase the existence probability. Similarly, when it is determined that there is no pattern corresponding to the object, the existence probability of the object corresponding to the pattern is set to the counterdown amount with 0% as a lower limit with respect to the existence probability calculated last time. Update by subtracting Ddn, and decrease the existence probability. In other words, according to the present embodiment, the presence state of the object is specified in stages in steps corresponding to the counter-up amount Dup and the counter-down amount Ddn as the existence probabilities.

  Here, in particular, when it is determined that the relative distance L is less than the threshold value Lth in the processing according to step A14, the counter up amount Dup and the counter down amount Ddn are each corrected to the decreasing side. In both the processes of C14 and step C15, the existence probability R (i) is relatively difficult to change (that is, compared with a case where the tunnel is not approaching). That is, according to the present embodiment, the counter up amount Dup and the counter down amount Ddn are each employed as an example of an index that defines the “reflection mode” according to the present invention, and the pattern detection accuracy of the pattern detection unit 130 is ensured. If it is determined that it is not performed, the detection result of the pattern is made difficult to be reflected in the identification of the existence state by correcting these to the decreasing side.

  The counter up amount Dup and the counter down amount Ddn may be corrected by holding the counter up amounts Dup1 and Dup2 and the counter down amounts Ddn1 and Ddn2 in the ROM or the like as fixed values and selecting one each time. Alternatively, correction amounts corresponding to “Dup1-Dup2” and “Ddn1-Ddn2” are stored in the ROM or the like in advance, and each time the correction amount is read and correction processing (ie, addition or subtraction) is performed. Also good.

  As described above, according to the third embodiment, the presence state of the object is specified stepwise by the existence probability indicating how much it can be regarded as existing, not the binary aspect of whether or not the object exists. Therefore, since the detection result of the past pattern can influence the identification of the present state of the target object to some extent, for some reason (for example, a physical, mechanical or electrical failure of the pattern detection unit 130, or Even if the pattern corresponding to the target object is temporarily not detected due to the behavior or state of the target object), the influence on the specific accuracy of the presence state is small. That is, according to this aspect, it is easy to ensure reliability with respect to detection of an object, which is advantageous in practice.

  Further, according to the present embodiment, the setting mode of the counter up amount Dup and the counter down amount Ddn (which is a fixed value in this embodiment, may be selectable by a driver, for example, or may be appropriately selected on the system. It is possible to change the base reflection mode (that is, the reflection mode in the normal state), depending on how many candidate values are obtained. For example, if the counter up amount Dup1 and the counter down amount Ddn1 are set relatively large, the latest pattern detection result is remarkably reflected in the identification of the presence state (that is, the sensitivity to the pattern detection result is high). On the other hand, if each is set relatively small, the latest pattern detection result is hardly reflected in the identification of the presence state (that is, the sensitivity to the pattern detection result is reduced). Therefore, it is possible to optimize the specific aspect of the existence probability in accordance with the type of the object.

  Note that the image quality of the obtained image is unlikely to be uniform between the tunnel entrance and the tunnel exit due to the physical characteristics of the image sensor 300. Therefore, the detection accuracy of the pattern detected by the pattern detection unit 130 can change depending on whether the pattern is approaching the entrance or the exit. In view of such circumstances, instead of performing a uniform correction when approaching the structure, an appropriate correction amount is set for each structure (may be prepared in advance) and pattern detection is performed. You may control more finely the reflection aspect at the time of reflecting a result on existence probability.

  In the above-described various embodiments, when the vehicle 10 is provided with a light source such as a near-infrared projector for emitting auxiliary light for assisting the imaging of the image sensor 300, it is related to the approach to the structure. Imaging of the image sensor 300 may be supported by auxiliary light from the light source. Although it is difficult to completely suppress the change in brightness caused by approaching the structure only by the action of such auxiliary light, it is possible to prevent deterioration of image quality to some extent, so the pattern detection accuracy is Can rise relatively. Therefore, the amount of decrease when correcting the counter-up amount Dup and the counter-down amount Ddn to the decrease side can be small, and the detection accuracy of the object can be brought closer to the normal time. Such auxiliary light may be continuously used inside the tunnel if the structure is a tunnel.

  In addition, the change in the reflection mode of the pattern detection result before and after approaching the tunnel entrance and exit described in each of the above embodiments is originally the brightness of the imaging region of the image sensor 300 before and after approaching the tunnel entrance and exit. This is done because the change is expected to be prominent, and even if it is the target structure, it is determined or predicted in advance that the change in brightness will not occur, or In the case where it can be sufficiently estimated, it is not necessary to change the reflection mode. For example, more specifically, since no significant change in brightness occurs inside and outside the tunnel at night, for example, a series of processing according to steps A13 to A16 in FIG. 2 is canceled and processing according to step A12 is performed. If it is determined that the pattern detected in step 1 is an object, the process according to step A17 may be immediately executed and extracted as an object.

  The present invention is not limited to the above-described embodiment, and can be appropriately changed without departing from the spirit or idea of the invention that can be read from the claims and the entire specification. Is also included in the technical scope of the present invention.

1 is a block diagram conceptually showing a configuration of a portion related to an object detection device according to the present invention in a vehicle according to a first embodiment of the present invention. It is a flowchart of the target object detection process performed by ECU in the vehicle of FIG. It is a flowchart of the target object detection process which concerns on 2nd Embodiment of this invention. It is a flowchart of the target object detection process which concerns on 3rd Embodiment of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... Vehicle, 100 ... ECU, 110 ... Control part, 120 ... Position information storage part, 130 ... Pattern detection part, 140 ... Object information storage part, 200 ... Car navigation apparatus, 300 ... Image sensor.

Claims (11)

An object detection device for detecting a predetermined type of object in a vehicle including an imaging unit capable of imaging a front detection range,
Detection means capable of detecting a pattern corresponding to the object in the image of the detection range obtained via the imaging means;
Presence state specifying means for specifying the presence state of the object in front based on the detection result of the pattern;
A positional relationship specification that specifies a relative positional relationship between the vehicle and a predetermined type of structure that exists in the travel route of the vehicle and is set in advance to increase the degree of change in brightness in the detection range. Means,
Discrimination means for discriminating whether or not the vehicle is approaching the structure based on the specified positional relationship;
Determining means for determining a reflection mode when reflecting the detection result of the pattern on the identification of the presence state based on the determination result; and
The presence state specifying means further specifies the presence state based on the determined reflection mode when the reflection mode is determined. The object detection apparatus according to claim 1, wherein the structure includes at least one of an entrance portion and an exit portion in a tunnel. The vehicle further includes position detection means for detecting absolute positions of the vehicle and the structure,
The target object detection apparatus according to claim 1, wherein the positional relation specifying unit specifies the positional relation based on the detected absolute position. The object detection apparatus according to any one of claims 1 to 3, wherein the existence state specifying unit specifies an existence probability of the object as the existence state. The positional relationship specifying means specifies a distance between the vehicle and the structure as the positional relationship,
The said discrimination | determination means performs the said discrimination | determination that it is approaching to the said structure when this specified distance is less than predetermined value. The Claim 1 characterized by the above-mentioned. Object detection device. The determination means determines the reflection mode so that the detection result of the pattern is less likely to be reflected in the identification of the existence state when the determination that the structure is approaching is performed. The target object detection device according to claim 1, wherein the target object detection device is a feature. The determination means determines the reflection mode so that the detection result of the pattern is not reflected in the specification of the existence state when it is determined that the structure is approaching. The object detection apparatus according to claim 6. The determination means may reflect the detection result of the pattern to the identification of the existence state when the same pattern is detected before and after the determination that the structure is approaching. The object detection apparatus according to claim 7, wherein the reflection mode is determined. The presence state specifying means, when it is determined that the structure is approaching, based on the presence state specified before the determination that the structure is approached The object detection apparatus according to claim 7 or 8, wherein the presence state is specified. The presence state specifying means specifies the presence probability of the object as the presence state, and increases the presence probability by a first correction amount each time the pattern is detected,
10. The determination unit according to claim 6, wherein the determination unit corrects the first correction amount to a decrease side when it is determined that the structure is approaching. 10. 2. The object detection apparatus described in 1. The presence state specifying means specifies the presence probability of the object as the presence state, and decreases the presence probability by a second correction amount every time the pattern is not detected,
The said determination means correct | amends a said 2nd correction amount to the reduction | decrease side, when discrimination | determination to the effect of approaching the said structure is performed. The any one of Claim 6 to 10 characterized by the above-mentioned. 2. The object detection apparatus described in 1.

Images