JP2009044631A - Object detection device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an object detection device for detecting an object by discriminating an object which is easily visible for a driver from an object which is hardly visible for the driver. <P>SOLUTION: An object detection device 1, provided with a high luminance region extraction part 13 for extracting a high luminance region K<SB>j</SB>having luminance which is larger than prescribed luminance; and a contact decision part 19 for deciding whether an object region O<SB>i</SB>corresponding to the object is brought into contact with the high luminance region K<SB>j</SB>, is configured to extract the high luminance region K<SB>j</SB>having luminance larger than the prescribed luminance from among pickup image information, decides whether the object region O<SB>i</SB>is brought into contact with the high luminance region K<SB>j</SB>, and detects only a pedestrian who does not carry any luminaire by discriminating whether a high luminance body is accompanied with the object. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an object detection device that detects an object.

2. Description of the Related Art Conventionally, an apparatus for detecting surrounding objects has been used for driving support for vehicles. For example, it is used in an apparatus that supports a driver's vision by visualizing a portion that is difficult for the driver to see at night. As such an apparatus, there is known an object detection apparatus that extracts a pedestrian from a captured image based on a shape and warns with an alarm sound, and emphasizes the extracted pedestrian and visualizes the image (for example, Patent Documents). 1).
JP 2005-159392 A

  However, in the conventional apparatus, warning processing is also performed on an object that can be visually recognized by the driver, and thus it may be troublesome for the driver to call attention. In addition, it is difficult to prioritize and emphasize objects that are difficult to see and need attention, such as bicycles that are not lit.

  Accordingly, the present invention has been made to solve such a technical problem, and provides an object detection device that performs object detection by identifying an object that is easy for the driver to see and an object that is difficult to see. With the goal.

  That is, an object detection device according to the present invention includes an object detection unit that detects an object from image information of an image captured by an on-vehicle imaging unit, and a high brightness that is greater than a predetermined luminance from the image information. A high-brightness body detection unit that extracts a brightness region and detects it as a high-brightness body, and based on the detection result of the object and the detection result of the high-brightness body, the object and the high brightness And an identification means for identifying the object not accompanied by a body.

  According to the present invention, a high-brightness area having a brightness greater than a predetermined brightness is extracted from the captured image information and detected as a high-brightness body, and whether or not the object is accompanied by a high-brightness body is identified. can do. As a result, it is possible to distinguish between an object accompanied by a high-luminance body, that is, an object that is relatively easy for the driver to visually recognize, and an object not accompanied by the high-luminance body, that is, an object that is relatively difficult to view.

  Here, the object detection device emphasizes the presence of the object not accompanied by the high-luminance body rather than the presence of the object accompanied by the high-luminance body, and notifies the driver to provide driving assistance for the vehicle. It is preferable to provide driving support means.

  By configuring in this way, the presence of an object not accompanied by a high-luminance body can be emphasized over the presence of an object accompanied by a high-luminance body and the driver can be notified to provide driving assistance for the vehicle. Therefore, the driver can quickly recognize the presence of an object that is relatively difficult to see.

  In addition, the object detection device may include driving support means that performs driving support by excluding the object accompanied by the high-luminance body from a target of driving support.

  By configuring in this way, an object accompanied by a high-luminance body can be excluded from the target of driving support, so the driver is not notified of the presence of an already recognized object. It can be avoided that the person feels bothersome.

  The object detection device preferably includes a light projecting unit that projects light in a predetermined light projecting direction toward the image capturing direction, and the image capturing unit preferably has a longer image capturing time than the light projecting time. .

  With this configuration, the imaging time of the imaging unit can be made longer than the projection time of the light projecting unit, so that an object that does not emit light as a still image at the timing when it is projected by the light projecting unit. An object that emits light can be captured as an image showing a trajectory of an operation during the imaging time. This makes it possible to distinguish between a high-luminance body that emits light and an object that is instantaneously reflected, so that it is possible to identify whether the high-luminance body is an illuminator.

  In addition, the object detection apparatus including the light projecting unit determines whether the feature amount deriving unit that derives the feature amount of the high brightness area, and whether the area corresponding to the object and the high brightness area are in contact with each other. If the contact determination unit determines that the region corresponding to the object and the high brightness region are in contact with each other, the high brightness body is an illuminator based on the feature amount. It is preferable to include an illuminator determination unit that determines whether or not.

  With this configuration, the feature amount of the high brightness area is derived, it is determined whether the area corresponding to the object and the high brightness area are in contact with each other, and if it is determined that they are in contact Based on the feature quantity of the high luminance area, it can be determined whether or not the high luminance body corresponding to the high luminance area is an illuminator. Thereby, it can be determined more accurately whether the high-luminance body is an illuminator.

  In the object detection apparatus including the light projecting unit, the feature amount deriving unit calculates an area of the high luminance region as the feature amount, and the illuminator determining unit is configured to calculate the region corresponding to the object and the high luminance region. It is preferable to determine whether or not the high-luminance body is an illuminator based on an area ratio with the region.

  By configuring in this way, the area of the high luminance area is calculated as the feature amount, and whether or not the high luminance body is an illuminator is determined based on the ratio of the area corresponding to the object and the area of the high luminance area. Can be determined. Thereby, it can be determined with higher accuracy whether or not the high-luminance body is an illuminator.

  Further, the object detection device comprising the light projecting means includes vanishing point determining means for determining the position of the vanishing point of the image based on the shape of the locus of the high luminance region, the position of the vanishing point and a predetermined pitch of the vehicle A pitch angle calculating unit that calculates the pitch angle of the vehicle by comparing the position of the vanishing point in the angle, and a range limiting unit that limits a search range of the image information based on the pitch angle. It is preferable.

  With this configuration, the position of the vanishing point of the image is determined based on the high luminance area, and the position of the vanishing point determined is compared with the position of the vanishing point at a predetermined pitch angle. Can be calculated. Further, since the search range of the image information can be limited based on the calculated pitch angle, the region corresponding to the object and the high luminance region can be extracted accurately and quickly.

  According to the present invention, an object can be detected by identifying an object that is easy for the driver to see and an object that is difficult to see.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same elements are denoted by the same reference numerals, and redundant description is omitted.

  First, the configuration of the object detection apparatus 1 according to the present embodiment will be described with reference to FIG. FIG. 1 is a configuration diagram of an object detection device 1 according to the present embodiment. The object detection device 1 according to the present embodiment is a device that detects a specific object from captured image information. For example, in driving a vehicle at night, driving that supports driving by detecting pedestrians and bicycles. It is used suitably for a support system.

  As shown in FIG. 1, the object detection apparatus 1 includes an information acquisition unit 50, an ECU 10, and an information provision unit 30. Here, the ECU (Electronic Control Unit) is a computer of an automobile device that is electronically controlled, and includes a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), an input / output interface, and the like. Configured.

  The information acquisition unit 50 has a function of acquiring information around the vehicle, and includes an imaging unit (imaging unit) 21, a projector (light projection unit) 22, and a vehicle speed sensor 53.

  The imaging unit 51 mainly has a function of acquiring image information in the vehicle traveling direction. The imaging unit 51 has a function of acquiring image information by changing the imaging time based on a signal input from the ECU 10. As the imaging unit 51, for example, a near infrared camera is used. The imaging unit 51 has a function of outputting captured image information to the ECU 10.

  The projector 52 has a function of projecting light in the direction in which the imaging unit 51 captures an image. The projector 52 has a function of changing the projection time based on a signal input from the ECU 10. As the projector 52, for example, a strobe that projects near infrared rays is used. Note that a projector that projects visible light, infrared light, or ultraviolet light may be used as the projector 52.

  The vehicle speed sensor 53 is a sensor that detects a wheel speed pulse. For example, the vehicle speed sensor 53 detects the rotation of the wheel by detecting a change in magnetic flux due to the rotation of the wheel, and acquires the speed of the vehicle. The vehicle speed sensor 53 has a function of outputting the acquired vehicle speed to the ECU 10.

  The ECU 10 includes a synchronization control unit 11, an information input unit 12, a high luminance region extraction unit (high luminance body detection unit) 13, a feature amount deriving unit (feature amount deriving unit) 14, and a vanishing point deriving unit (vanishing point determining unit) 15. , Pitch angle calculation unit (pitch angle calculation unit) 16, search range limitation unit (range limitation unit) 17, object region extraction unit (object detection unit) 18, contact determination unit (contact determination unit, identification unit) 19, illuminator A determination unit (illuminator determination unit) 20, an object identification unit (identification unit) 21, and an output unit (driving support unit) 22 are configured.

  The synchronization control unit 11 has a function of controlling the image capturing time of the image capturing unit 51 and the light projecting time of the projector 52. The synchronization control unit 11 outputs a synchronization signal to the imaging unit 51 and the projector 52 to control the imaging time and the light projection time. The synchronization control unit 11 has a function of inputting the vehicle speed from the vehicle speed sensor 53 and controlling the imaging time and the light projection time based on the input vehicle speed. Note that the function of synchronous control based on the vehicle speed is not necessarily a necessary function, and may be provided in response to a request for each vehicle. Further, the synchronization control unit 11 has a function of outputting the input timing of image information captured by the imaging unit 51 to the information input unit 12.

  The information input unit 12 has a function of inputting the image information captured by the imaging unit 51 based on the input timing of the image information input from the synchronization control unit 11. In addition, the information input unit 12 has a function of outputting the input image information to the high luminance region extraction unit 13.

  The high brightness area extraction unit 13 has a function of inputting the brightness of the image information and extracting pixels having brightness greater than a predetermined brightness. The high luminance area extraction unit 13 determines whether or not each pixel of the image information input from the information input unit 12 is larger than a predetermined luminance, and if it is determined that the pixel is larger than the predetermined luminance, It has a function of assigning a high brightness flag indicating high brightness and extracting only information related to the high brightness area from the image information based on the assigned high brightness flag. For example, the high luminance flag is assigned by changing the information bit from 0 to 1 in a predetermined memory area. The high brightness area extracting unit 13 has a function of outputting image information to the search range limiting unit 17 and outputting information about the extracted high brightness area to the feature amount deriving unit 14.

  The feature amount deriving unit 14 has a function of receiving information on the high luminance region from the high luminance region extracting unit 13 and deriving the feature amount of the input high luminance region. The feature amount deriving unit 14 derives the inertia main axis of the high luminance region, and calculates the area of the high luminance region, the area centroid position, the long axis length of the inertia main axis, the angle with respect to the image horizontal line, the short axis length of the inertia main axis, the circumference length, and the like. Derived as feature quantity. The feature quantity deriving unit 14 has a function of outputting the derived feature quantity to the vanishing point deriving unit 15.

  The vanishing point deriving unit 15 has a function of deriving the vanishing point of the image based on the feature amount input from the feature amount deriving unit 14. The vanishing point deriving unit 15 calculates a straight line passing through the major axis of the inertial main axis of the high-luminance region for each high-luminance region, and derives the position of the point where each straight line intersects (the position of the maximum likelihood intersection). As a method for calculating the position of the maximum likelihood intersection, for example, a voting method is used. The calculated maximum likelihood intersection is the vanishing point of the image. The vanishing point deriving unit 15 has a function of outputting the derived position of the maximum likelihood intersection to the pitch angle deriving unit 16.

  The pitch angle calculation unit 16 has a function of calculating the pitch angle of the vehicle using the maximum likelihood intersection (vanishing point) input from the vanishing point deriving unit 15. The pitch angle calculation unit 16 calculates the pitch angle of the vehicle based on the difference between the position of the maximum likelihood intersection and the position of the vanishing point of the image captured at the predetermined pitch angle. The pitch angle deriving unit 16 has a function of outputting the calculated pitch angle to the search range limiting unit 17.

  The search range limiting unit 17 estimates the position of the road plane based on the image information input from the high luminance region extraction unit 13 and the pitch angle input from the pitch angle calculation unit 16, and performs object detection processing such as a pedestrian. It has a function to set a search range. In addition, the search range limiting unit 17 has a function of outputting image information, the feature amount of the high luminance region, and information regarding the limited search range to the object region extraction unit 18.

  The object area extraction unit 18 has a function of extracting an area corresponding to an object such as a pedestrian, that is, an object area, from the image range limited by the search range limitation unit 17. For object region extraction, for example, template matching using normalized correlation, neural network, or image recognition technology such as SVM (Support Vector Machine) is used. In addition, the object area extraction unit 18 creates detected object information such as an ID number, position information, vertical width information, horizontal width information, and attribute information corresponding to each of the extracted object areas. Further, the object area extraction unit 18 has a function of outputting image information, feature amounts of the high luminance area, and information on the object area to the contact determination unit 19.

  The contact determination unit 19 determines whether the end of the high luminance region on the maximum likelihood intersection side and the object region are in contact with each other based on the feature amount of the high luminance region input from the object region extraction unit 18 and information on the object region. It has a function to determine whether. Further, the contact determination unit 19 has a function of outputting image information and a determination result to the illuminator determination unit 20.

  The illuminator determination unit 20 has a function of determining whether or not the high brightness area is an illuminator based on the determination result input from the contact determination unit 19. The illuminator determination unit 20 determines the area ratio between the contacting object and the high luminance region, the complexity of the high luminance region, the inertial main axis (for the high luminance region determined to be in contact with the high luminance region and the object region. It has a function of determining the illuminator-likeness using the condition of the direction of the long axis. The illuminator determination unit 20 has a function of outputting image information and a determination result to the object identification unit 21.

  Based on the determination result input from the illuminator determination unit 20, the object identification unit 21 has a function of identifying an object that has an illuminator and an object that does not have an illuminator. Moreover, the object identification part 21 has a function which excludes the detection object information regarding the object which has an illuminator. The method of exclusion may be a method of deleting the detected object information itself relating to the object having the illuminator from the detected object information, or a method of recording the invalid information in the attribute information of the detected object information. There may be. The object identification unit 21 has a function of outputting image information and identified detected object information to the output unit 22.

  The output unit 22 has a function of outputting a driving support command to the information providing unit 30 based on the image information input from the object identifying unit 21 and the detected detected object information.

  The information providing unit 30 has a function of providing information to the driver based on information obtained from the ECU 10, and includes, for example, a display 31 and a speaker 32.

  Next, operation | movement of the object detection apparatus 1 which concerns on this embodiment is demonstrated using FIGS. 2 to 5 are flowcharts illustrating the operation of the object detection apparatus 1 according to the present embodiment, FIG. 6 is a schematic diagram illustrating the relationship between the projection time and the imaging time, and FIG. 7 illustrates the leading curtain synchronization and the slow synchronization. Examples of images when used, FIGS. 8 to 17, are schematic diagrams for explaining the operation of the object detection apparatus 1 according to the present embodiment.

  First, the operation of the object detection apparatus 1 will be described with reference to FIGS. 2 and 3. 2 and 3, for example, a vehicle is provided with a driving support button, and is started at a timing when the driving support button is turned on, and is repeatedly executed at a predetermined interval.

  The object detection device 1 starts from the vehicle speed input process (S10). The process of S10 is executed by the vehicle speed sensor 53 and the synchronization control unit 11, and is a process of inputting the vehicle speed to the ECU 10. When the process of S10 ends, the process proceeds to a light projection pulse output process (S12).

  The process of S <b> 12 is a process that is executed by the synchronization control unit 11 and the projector 52, and outputs a projection pulse for causing the projector 52 to emit light. For example, as shown in FIG. 6B, the synchronization control unit 11 emits light when the projector 52 is in the ON state and does not emit light when the projector 52 is in the OFF state. Outputs a pulse repeated at. The time length of the pulse at which the light projector 52 is turned on, that is, the repetitive light projecting time is shorter than the repetitive image capturing time of the image capturing unit 51, and is 5 msec here. When the processing of S12 ends, the process proceeds to image capturing processing (S14).

  The process of S14 is executed by the synchronization control unit 11 and the imaging unit 51, determines the imaging time length, and captures an image. For example, as illustrated in FIG. 6A, the synchronization control unit 11 starts imaging of the imaging unit 51 when the imaging button of the imaging unit 51 is in the ON state, and when the imaging button of the imaging unit 51 is in the OFF state. When the imaging of the imaging unit 51 is finished, a pulse in which the ON state and the OFF state are repeated at a predetermined interval is output. Further, the time length during which the image pickup button of the image pickup unit 51 is turned on, that is, the image pickup time length is longer than the light projection time of the projector 52, and is so-called slow sync. Here, the imaging time length is 100 msec. Furthermore, the synchronization control unit 11 controls the imaging time start timing to be synchronized with the light projection start timing of the projector 52. In this way, in the case of performing so-called front curtain synchronization in which imaging is started in synchronization with the start timing of the imaging time and the projection timing, if the imaging time is sufficiently longer than the projection time, as shown in FIG. It is possible to capture an image obtained by combining a still image captured by the reflected light of the light projected by the projector 52 and an image indicating a trajectory of an object that emits light during the imaging time. In FIG. 7, for example, the motorcycle 40, the occupant 41, the tree 42, and the like are still images at the positions projected by the projector 52, and the headlight 43 of the motorcycle 40 is a trajectory 44 that has moved during the imaging time. It is captured and displayed. As described above, in the process of S <b> 14, the synchronization control unit 11 controls the projector 52 and the imaging unit 51 to capture an image, and outputs the captured image information to the information input unit 12. When the process of S14 ends, the process proceeds to a luminance value determination process (S16).

  The process of S16 is a process executed by the high brightness area extraction unit 13 to determine the brightness value of the input image. The high luminance area extracting unit 13 inputs pixel information of the input image information and determines whether or not the pixel information is larger than a predetermined luminance value. The input of the pixel information is executed, for example, from the smallest coordinate as the position information. As an example of the predetermined luminance value, 230 is set here out of 256 gradations of luminance values 0 to 255. If it is determined in step S16 that the luminance value is greater than the predetermined luminance value, the process proceeds to a high luminance flag applying process (S18). On the other hand, if it is determined in step S16 that the luminance value is not greater than the predetermined luminance value, the process proceeds to the execution determination process for the entire range (S20).

  The process of S18 is a process that is executed by the high-intensity region extraction unit 13 and adds a high-intensity flag indicating that the pixel information determined in the process of S16 is high-intensity to the determined pixel information. The high-intensity flag is assigned, for example, by changing the information bit from 0 to 1 in a predetermined memory area. When the process of S18 ends, the process proceeds to the full range execution determination process (S20).

  The process of S20 is a process of determining whether or not the process of S16 has been executed for the pixel information of the entire range of the image. In the process of S20, when the pixel information of the entire range has not been determined, the process proceeds to S16 again. On the other hand, in the process of S20, when the pixel information of the entire range is determined, the process proceeds to a high luminance pixel extraction process (S22).

  The process of S22 is a process that is executed by the high brightness area extraction unit 13 and extracts only the pixels to which the high brightness flag is given in the process of S18. For example, if the image input in the process of S14 is the image shown in FIG. 8, the image obtained by extracting only the pixels corresponding to the high luminance area is the image shown in FIG. When the process of S22 ends, the process proceeds to the ID assigning process for the high luminance area (S24).

The process of S24 is a process that is executed by the high-intensity area extraction unit 13 and assigns an ID to each of the high-intensity areas configured by the pixels extracted in the process of S22. The high luminance area extraction unit 13 confirms the connectivity of the pixels extracted in the process of S22, and assigns an ID for each high luminance area as shown in FIG. The ID is, for example, a number, and the high luminance area is identified by using the assigned ID, for example, K _j (j: natural number). When the process of S24 ends, the process proceeds to a feature amount derivation process (S26).

The process of S26 is a process that is executed by the feature quantity deriving unit 14 and derives the feature quantity of each high-brightness region K _j to which the ID is assigned in the process of S24. The feature amount of the high luminance region K _j will be described with reference to FIG. Figure 11 is a schematic diagram illustrating the features of the high luminance region K ₁ in FIG. 10, the feature size of the high luminance region is information including orientation. For example, the area _{S 1} of the high luminance region _{K 1} as feature amounts, centroid position _{(X 1, Y 1),} the major axis length _{L 1} of the principal axis of inertia, the angle alpha ₁ relative to the image horizontal line, the short axis length of the principal axis of inertia _{M 1} The information including the peripheral length C _{1 and the} like is derived. Feature amount derivation unit 14, for each of the high luminance region _{K j,} the area _{S j} of the high luminance region _{K j,} centroid position _{(X j, Y j),} the length of the long axis of the principal axes of inertia _{L j,} the angle with respect to the image horizontal line α _j , the short axis length M _j of the inertial main axis, the perimeter length C _{j and the} like are derived as feature quantities. When the process of S26 ends, the process proceeds to the maximum likelihood intersection derivation process (S28).

Processing of S28 is executed by the vanishing point derivation unit 15 and deriving a vanishing point based on the feature amount of the high luminance region K _j derived in the processing of S26. As shown in FIG. 12, the vanishing point deriving unit 15 derives the maximum likelihood intersection P _f that intersects with the extension of the principal axis of inertia of each high-luminance region K _j . This maximum likelihood intersection P _f is determined by a voting method or the like. Further, the derived maximum likelihood intersection point P _f becomes a vanishing point of the image. When the process of S28 is completed, the process proceeds to a pitch angle derivation process (S30).

Processing step S30 is performed by the pitch angle calculator 16, a process of calculating the pitch angle of the vehicle from the maximum likelihood intersection (vanishing point) P _f derived in S28. The pitch angle at the maximum likelihood intersection P _f is β, and as shown in FIG. 13, the Y-axis coordinate value of the vanishing point is Y _b , and the Y coordinate value of the vanishing point when the pitch angle is 0 ° is Y ₀ . Then, the pitch angle β of the host vehicle can be calculated by Expression (1) using the vertical viewing angle Θ of the image and the vertical pixel number n.

β = (Y _b −Y ₀ ) · Θ / n (1)

  When the processing of S30 ends, the process proceeds to search range limitation processing (S32).

  The process of S32 is executed by the search range limiting unit 17 and is a process of estimating the position of the road plane based on the pitch angle β calculated in the process of S30 and limiting the image range for searching for an object. For example, as shown in FIG. 14, the search range limiting unit 17 shifts the search range upward by the pitch angle β with reference to the search range when the pitch angle is 0 °. When the process of S32 ends, the process proceeds to an object region extraction process (S34).

The process of S34 is performed by the object area extraction unit 18 and extracts an object area corresponding to an object within the search range of the image obtained by the process of S32. Here, the detected object is a pedestrian. The object region extraction unit 18 detects an image region corresponding to a pedestrian as shown in FIG. 15 using image recognition technology such as template matching, neural network, or SVM. The object region extraction unit 18 extracts an image region corresponding to the detected pedestrian as, for example, a rectangular object region, and creates detected object information for each. For example, when the object area O ₀ corresponding to the pedestrian shown in FIG. 15 is detected, ID = 0, position = (x ₀ , y ₀ ), and vertical width = H as detected object information of the object area O _{0. 0} , width = W ₀ , attribute = PPP are created. Also, when detecting an object area _{O 1} corresponding to a pedestrian with a bicycle, as the detection object information of the object area _{O 1,} ID = 1, the position _{= (x 1, y 1)} , the vertical width = _{H 1} , Width = W ₁ , attribute = PPP. As the attribute information, a character string indicating the type of the detected object and the validity of the information is adopted. Here, PPP is an attribute indicating a pedestrian. As described above, the object region extraction unit 18 extracts a region corresponding to the object, and creates detected object information for each extracted region. When the process of S34 ends, the process proceeds to the illuminator possession determination process (S36).

The process of S36 is a process that is executed by the contact determination unit 19 and the illuminator determination unit 20, and determines whether or not the high luminance region _Kj is an illuminator. Details of this processing will be described later. When the process of S36 is completed, the process proceeds to an object identification process (S38).

  The process of S38 is a process executed by the object identification unit 21 to exclude an object having an illuminator from the object information from the detected object information. Details of this processing will be described later. When the process of S38 is completed, the process proceeds to an information output process (S40).

  The process of S40 is a process that is executed by the output unit 22 and outputs the detected object information obtained by the process of S38 to assist driving. For example, the output unit 22 emphasizes a detected object that does not have an illuminator in the image information displayed on the display 31 or alerts the driver using the speaker 32.

  As described above, the object detection device 1 efficiently detects an object by limiting a search range using an image in which an imaging time and a light projection time are controlled, and detects a pedestrian who does not have an illuminator. It is possible to identify and provide driving assistance.

  Next, the illuminator possession determination process will be described in detail with reference to FIG. FIG. 4 is a flowchart showing details of the process of S36 of FIG. The illuminator possession determination process is started from an initialization process (S50).

  The process of S50 is executed by the contact determination unit 19 and initializes variables. Here, since variables i (i: integer) and j are used, variable i is set to 0 and variable j is set to 1. When the process of S50 ends, the process proceeds to a contact determination process (S52).

The process of S52 is executed by the contact determination unit 19 and determines whether or not the light intensity area K _j extracted in S22 of FIG. 2 is associated with the object area O _i extracted in S34 of FIG. . The determination as to whether or not it accompanies is made based on whether or not the end of the light luminance area K _j close to the maximum likelihood intersection P _f and the object area O _i are in contact with each other. Determination of the contact is not only in contact adjacent the object region O _i and the high luminance region K _j, the end is present close to the maximum likelihood intersection P _f of light intensity region K _j in the object area O _i Also in this case, it is determined that the end of the light luminance area K _j close to the maximum likelihood intersection P _f is in contact with the object area O _i . For example, as shown in FIG. 16, when the object area O ₁ and the light intensity area K ₇ are determined, the end of the light intensity area K _{7 on} the maximum likelihood intersection P _f side exists in the object area O ₁ . It is determined that the object region O ₁ and the light luminance region K ₇ are in contact with each other. In the processing of S52, when the end to the object area O _i close to the maximum likelihood intersection P _f of light intensity region K _j is determined that contact, the process proceeds to the illuminator determination process (S54).

The process of S54 is executed by the illuminator determination unit 20, and is a process of determining whether or not the high luminance region _Kj is an illuminator. The determination as to whether or not it is an illuminator is made based on the area ratio between the high brightness area K _j and the object area O _i determined to be in contact in the process of S52. For example, if the area of the high brightness region K _j is S _j , the area of the object region O _i is S _i , the minimum value of the assumed area ratio is Ts ₁ , and the maximum value of the assumed area ratio is Ts ₂ , It is determined whether or not the condition (2) is satisfied.

Ts ₁ <(S _j / S _i ) <Ts ₂ (2)

Ts ₁ and Ts ₂ are set values. For example, Ts ₁ = 0.1 and Ts ₂ = 0.2 are set. In the process of S54, when it determines with it being an illuminator, it transfers to a possession flag provision process (S56).

The process of S56 is executed by the illuminator determination unit 20, and is a process of assigning an illuminator possession flag to the object region O _i determined to possess the illuminator in the process of S54. For example, the illuminator possession flag is assigned by changing the information bit from 0 to 1 in a predetermined memory area. When the process of S56 is completed, the process proceeds to an execution determination process for all object regions O _i (S60).

The process of S60 is executed by the illuminator determination unit 20, and is a process of determining whether or not the illuminator determination process of S54 has been executed for all object regions O _i . In the process of S60, when the illuminator determination process is performed for all the object regions O _i , the control process of FIG. 4 is terminated.

On the other hand, in the processing of S52, when the end to the object area O _i close to the maximum likelihood intersection P _f of light intensity region K _j is determined that no contact is execution determination processing of all the high luminance region K _j (S58). Processing step S58 is performed by the illuminator determination unit 20, is a process of determining whether to execute the contact determination process of S52 with respect to all the light intensity region K _j. In the processing of S58, when you have made contact determination process for all the light intensity region K _j, the process proceeds to execution determination processing of all the object areas O _i (S60). Further, in the process of S58, if the system is not contact determination process for all of the high luminance region K _j, the process proceeds to the judgment target changing process (S62).

The process of S62 is a process that is executed by the illuminator determination unit 20 and changes the high-intensity region _Kj that is a determination target of the contact determination process of S52. The illuminator determination unit 20 sets j to j + 1, and changes the determination target from the light luminance region K _j to the light luminance region K _{j + 1} . When the process of S62 ends, the process proceeds to a contact determination process (S52).

On the other hand, in the process of S60, when the illuminator determination process is not performed for all the object regions O _i , the process proceeds to the determination target change process (S64). The process of S64 is a process that is executed by the illuminator determination unit 20 and changes the object region O _i that is the determination target of the contact determination process of S52. The illuminator determination unit 20 sets i to i + 1, and changes the determination target from the object region O _i to the object region O _{i + 1} . When the process of S62 ends, the process proceeds to a contact determination process (S52).

  Thus, the control process of FIG. 4 ends. With the control process of FIG. 4, it can be determined from the captured image whether the object has an illuminator.

  Next, the object identification process will be described in detail with reference to FIG. FIG. 5 is a flowchart showing details of the process in S38 of FIG. The object identification process is started from the initialization process (S70).

  The process of S70 is executed by the object identification unit 21 and is a process for initializing variables. Here, since the variable i is used, the variable i is set to 0. When the processing of S70 ends, the process proceeds to possession determination processing (S72).

The process of S72 is a process that is executed by the object identification unit 21 and determines whether or not the possession flag is assigned to the object region O _i . In the process of S72, when it is determined that the possession flag is given, the process proceeds to an exclusion process (S74). On the other hand, in the process of S72, when it is determined that the possession flag is not given, the process proceeds to the execution determination process for all the object areas O _i (S76).

The process of S74 is executed by the object identification unit 21, and the object region O _i determined to have the possession flag given in the process of S72, that is, the information related to the object determined to have the illuminator It is a process of deleting from the information. When the process of S74 is completed, the process proceeds to an execution determination process for all object regions O _i (S76).

The process of S76 is executed by the object identification unit 21, and is a process of determining whether or not the possession determination process of S72 has been performed for all object regions. In the process of S76, when it is determined that the possession determination process has been performed for all the object areas O _i , the control process of FIG. 5 is terminated. On the other hand, in the process of S76, when it is determined that the possession determination process is not performed for all the object regions O _i , the process proceeds to the determination target change process (S78).

The process of S78 is a process that is executed by the object identification unit 21 and changes the object region O _i to be determined. The illuminator determination unit 20 performs i = _{i + 1} and changes the determination target from the object region O _i to the object region O _{i + 1} . When the processing of S78 ends, the process proceeds to possession determination processing (S72).

As described above, by performing the control process of FIG. 5, an object having an illuminator and an object not having an illuminator can be identified, and the object having an illuminator can be excluded from driving support. For example, as shown in FIG. 17, only an object corresponding to the object region O ₀ that does not have an illuminator can be set as a driving assistance target.

As described above, the object detection apparatus 1 according to the present embodiment extracts a high-luminance region K _j having a luminance greater than a predetermined luminance from the captured image information, detects it as a high-luminance body, and supports driving. Identifying whether or not a high-luminance body is attached to the object by determining whether or not the object region O _i corresponding to the target candidate object and the high-luminance region K _j are in contact with each other. Can do. As a result, it is possible to distinguish between an object accompanied by a high-luminance body, that is, an object that is relatively easy for the driver to visually recognize, and an object not accompanied by the high-luminance body, that is, an object that is relatively difficult to view. Therefore, it is possible to detect only objects that are relatively difficult for the driver to see.

  Further, the object detection device 1 according to the present embodiment emphasizes the presence of an object not accompanied by a high-luminance body over the presence of an object accompanied by a high-luminance body, and notifies the driver of the vehicle. Since driving assistance can be performed, it is possible to quickly recognize an object that is relatively difficult for the driver to see.

  Moreover, since the object detection apparatus 1 according to the present embodiment can exclude an object accompanied by a high-luminance body from the target of driving support, the driver is notified of the presence of an already recognized object. Therefore, it can be avoided that the driver feels bothersome.

  Moreover, since the object detection apparatus 1 according to the present embodiment can make the imaging time of the imaging unit 51 longer than the light projection time of the projector 52, an object that does not emit light by itself is projected by the light projecting means. As a still image of timing, an object that emits light by itself can be captured as an image showing a trajectory of an operation during the imaging time. This makes it possible to distinguish between a high-luminance body that emits light and an object that is instantaneously reflected, so that it is possible to identify whether the high-luminance body is an illuminator.

Further, the object detecting device 1 according to this embodiment calculates a feature quantity of the high luminance region K _j, determines whether or not the object region O _i and the high luminance region K _j are in contact with each other, in contact with If it is determined that can be based on the feature amount of the high luminance region K _j, high brightness bodies corresponding to the high luminance region K _j to determine whether the illuminator. Thereby, it can be determined more accurately whether the high-luminance body is an illuminator.

The ratio of the object detection apparatus 1 according to this embodiment, the area of high as the feature quantity of luminance region K _j calculates the area of the high luminance region K _j, object area O _i of the area and the high luminance region K _j Based on the above, it can be determined whether or not the high-luminance body is an illuminator. Thereby, it can be determined with higher accuracy whether or not the high-luminance body is an illuminator.

Furthermore, the object detection apparatus 1 according to the present embodiment determines the position of the vanishing point of the image based on the high luminance region K _j , and determines the position of the vanishing point determined and the position of the vanishing point at a predetermined pitch angle. In comparison, the pitch angle β of the vehicle can be calculated. Further, since the search range of the image information can be limited based on the calculated pitch angle β, the object region O _i and the high luminance region K _j can be accurately and quickly extracted.

  The above-described embodiment shows an example of the object detection apparatus according to the present invention. The object detection device according to the present invention is not limited to the above, and the object detection device according to the embodiment is modified or applied to others so as not to change the gist described in each claim. It may be a thing.

For example, in the embodiment, the case where the synchronization between the imaging unit 51 and the projector 52 is the first curtain sync, but the second curtain sync may be used. In the case of the trailing curtain sync, it is possible to obtain the same effect as that of the present embodiment by determining whether or not the end of the high luminance area K _j opposite to the maximum likelihood intersection point P _f is in contact with the object area O _i. Can do.

  Further, in the embodiment, an example using an image captured by slow sync as an optimal example of the object detection device has been described. However, even when an image captured by a normal imaging method is used, Similar effects can be obtained.

Further, in the embodiment, depending on whether it contains a near end to the maximum likelihood intersection P _f of the high luminance region K _j in the object area O _i, or the high luminance region K _j and the object region O _i contacts whether However, for example, it is determined whether there is an end close to the maximum likelihood intersection P _f of the high luminance region K _j within a set distance from the center position of the object region O _i. It may be determined whether or not they are in contact with each other based on the positional relationship between the region O _i and the end of the high luminance region K _j close to the maximum likelihood intersection P _f .

Further, in the embodiment, the example in which the determination as to whether or not it is an illuminator has been described based on the area ratio between the high luminance region K _j and the object region O _i , but for example, the complexity of the high luminance region K _j It may be determined using. Complexity A high-brightness region _{K j} is a perimeter _{C j,} and the area and ^{_{S j, A = C j 2}} / can be represented by _{S j,} if complexity is greater than a predetermined value, high It can be determined that the luminance region K _j is not an illuminator.

In the embodiment, the example in which the determination as to whether or not the illuminator is an illuminator has been described based on the area ratio between the high luminance region K _j and the object region O _i , but the direction of the principal axis of inertia of the high luminance region K _j is described. You may determine whether it is an illuminator using. For example, a case where the direction of the principal axis of inertia of the high-luminance region K _j is nearly vertical can be excluded as not satisfying the condition of the illuminator.

In the embodiment, the example in which the object region O _i holding the illuminator is deleted as a method of excluding from the detected object information has been described. However, for example, the attribute information may be changed. In this case, it is assumed that the detected object information is invalid when attribute information = NNN, and in the process of S74 in FIG. 5, the object region O _i that holds the illuminator is not deleted from the detected object information. the attribute information of the detected object information of the object area O _i that hold vessel by changing from PPP to NNN, it is possible to exclude the object area O _i that hold the illuminator from the detection object information.

Further, in the embodiment, without excluding the object area O _i that hold illuminator has been described how to exclude from detection object information, the object area O _i that hold the illuminator from the detection object information It is also possible to change the alert level. For example, when the attribute information is LOW, it is assumed that the alert level of the detected object information is low, and the object region O _i holding the illuminator is deleted from the detected object information in the process of S74 of FIG. Without changing the attribute information of the detected object information of the object region O _i that holds the illuminator from PPP to LOW, the object that holds the illuminator has a warning level among the detected object information. Information can be provided to the driver as being low. The case where the alerting level is low is realized, for example, by thinly displaying an object holding the illuminator on the display or lowering the warning volume on the speaker.

It is a block diagram of the object detection apparatus which concerns on this embodiment. It is a flowchart which shows operation | movement of the object detection apparatus of FIG. It is a flowchart which shows operation | movement of the object detection apparatus of FIG. It is a flowchart which shows the detail of the illuminator possession judgment process of FIG. FIG. 4 is a flowchart showing details of an object identification process of FIG. 3. FIG. It is a schematic diagram which shows the imaging time of an imaging part, and the light projection time of a light projector. It is the example of an image imaged with front curtain sync and slow sync. It is the example of an image imaged with front curtain sync and slow sync. FIG. 9 is an image example in which only a high luminance region is extracted from the image of FIG. 8. FIG. FIG. 10 is an example of an image in which an ID is assigned to the high luminance region shown in FIG. 9. FIG. It is a schematic diagram which shows the feature-value of the high-intensity area | region shown in FIG. It is an example of an image for explaining derivation of the maximum likelihood intersection. It is an example of an image for explaining operation of a pitch angle calculation part. It is an example of an image for explaining operation which limits a search range. It is an example of an image for explaining extraction of an object area. It is an example of an image for demonstrating the contact determination of a high-intensity area | region and an object area | region. It is an example of an image for explaining identification of an object field.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Object detection apparatus, 13 ... High-intensity area | region extraction part (high-intensity body detection means), 14 ... Feature-value derivation | leading-out part (feature-value derivation means), 15 ... Vanishing point derivation | leading-out part (vanishing point determination means), 16 ... Pitch Angle calculation unit (pitch angle calculation unit), 17 ... Search range limitation unit (range limitation unit), 18 ... Object region extraction unit (object detection unit), 19 ... Contact determination unit (identification unit, contact determination unit), 20 ... Illuminator determination unit (illuminator determination unit), 21... Object identification unit (identification unit), 22... Output unit (driving support unit).

Claims (7)

Object detection means for detecting an object from image information of an image captured by an on-vehicle imaging means;
A high-luminance body detection means for extracting a high-luminance region having a luminance greater than a predetermined luminance from the image information and detecting the high-luminance body;
Identification means for identifying the object accompanied by the high-luminance body and the object not accompanied by the high-luminance body based on the detection result of the object and the detection result of the high-luminance body;
An object detection apparatus comprising:   Driving support means for supporting the driving of the vehicle by notifying the driver by emphasizing the presence of the object not accompanied by the high-luminance body rather than the presence of the object accompanied by the high-luminance body. The object detection apparatus according to claim 1.   The object detection apparatus according to claim 1, further comprising driving support means that performs driving support by excluding the object accompanied by the high-luminance body from a driving support target. Comprising light projecting means for projecting in a predetermined projecting time toward the imaging direction of the image,
The imaging means makes the imaging time longer than the projection time;
The object detection device according to claim 1, wherein Feature quantity deriving means for deriving a feature quantity of the high luminance region;
Contact determination means for determining whether or not the region corresponding to the object and the high-intensity region are in contact with each other;
If it is determined by the contact determination means that the region corresponding to the object and the high luminance region are in contact with each other, whether or not the high luminance body is an illuminator is determined based on the feature amount. Illuminator determination means for determining;
The object detection apparatus according to claim 4, further comprising: The feature amount deriving unit calculates an area of the high luminance region as the feature amount,
The said illuminator determination means determines whether the said high-intensity body is an illuminator based on the area ratio of the area | region corresponding to the said object, and the said high-intensity area | region. The object detection apparatus described. Vanishing point determining means for determining the position of the vanishing point of the image based on the shape of the locus of the high-luminance region;
A pitch angle calculating means for calculating the pitch angle of the vehicle by comparing the position of the vanishing point and the position of the vanishing point at a predetermined pitch angle of the vehicle;
A range limiting means for limiting a search range of the image information based on the pitch angle;
The object detection device according to claim 4, wherein

Images