JP2013228334A - Three-dimensional measuring system, three-dimensional measuring method and three-dimensional measuring program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a noncontact three-dimensional measuring system capable of automatically recognizing a region of a human body and of measuring a dimension of the recognized region.SOLUTION: A three-dimensional measuring system 1 includes: multiple imaging units 16 to 19 for performing stereo photography; and a calculation controlling device 22 for generating a three-dimensional image. The calculation controlling device includes: a first determination unit 26 that generates three-dimensional shape data for a measurement object by a TIN based on feature points, sets horizontal sections at predetermined intervals in a vertical direction in the three-dimensional shape data, generates section lines by connecting section points meaning intersection points between the horizontal sections and a triangle constituting the TIN, and generates a section line group by repeatedly putting one section line on top of another at prescribed intervals; and a second determination unit 27 that obtains a centroid axis on the basis of an average centroid position of the section lines, specifies each point of the section lines in a prescribed angular direction around the centroid position as a set section point, sets a longitudinal section including the set section point and the centroid axis, and determines a position of a prescribed region on the basis of an external body shape line formed by connecting the set section points within the longitudinal section.

Description

  The present invention relates to a three-dimensional measurement system, a three-dimensional measurement method, and a three-dimensional measurement program, in particular, a three-dimensional measurement system, a three-dimensional measurement method, and a three-dimensional measurement method that measure a shape of a measurement object such as a human body in a non-contact manner. It relates to measurement programs.

  When measuring the three-dimensional shape of a measurement object such as a human body, the three-dimensional measurement is performed by stereo imaging with two cameras. By performing the three-dimensional measurement by stereo imaging, the three-dimensional shape of the measurement object can be quickly measured without contact.

  In recent years, it has been demanded to realize measurement of a three-dimensional shape of a human body by stereo photography in order to quickly and non-contactly measure the size of a garment that matches the body shape of a garment purchaser.

  In the case of the non-contact 3D measurement device as described above, noise may be mixed in the 3D data of the measured object due to noise light, camera performance, etc. It was not possible to perform this, and it was difficult to obtain an accurate numerical value for specifying the three-dimensional shape of the measurement object.

  In addition, the stereo image is taken using the feature points projected onto the measurement object by the feature projection unit, and the miscorresponding points in the duplicate image are automatically determined and removed, and then the three-dimensional shape of the measurement object is removed. As a shape measuring apparatus and a program for measuring the above, there is one disclosed in Patent Document 1.

JP 2010-133751 A

  In view of such circumstances, the present invention is a non-contact three-dimensional measurement system, a three-dimensional measurement method, and a three-dimensional measurement program capable of automatically recognizing a part of a human body from a stereo image and measuring the dimension of the recognized part. Is to provide.

  The present invention extracts a feature point from a plurality of image pickup units having two cameras for performing stereo shooting of a measurement object, and two images obtained by the image pickup unit, and corresponds to a common feature point on the image A three-dimensional measurement system having a three-dimensional measurement as a point, and measuring and calculating a three-dimensional shape of the measurement object for each imaging unit and creating a three-dimensional image of the measurement object, A plurality of imaging units are arranged to image the measurement object from a plurality of directions, and the arithmetic and control unit creates three-dimensional shape data of the measurement object by TIN created based on the feature points, A horizontal section is set in the three-dimensional shape data at predetermined intervals in the vertical direction, and a section line is created by connecting section points that are intersections of the horizontal section and the triangles forming the TIN, and the section line is defined as a predetermined section line. A first determination unit that creates a group of cross-sectional lines by laminating at intervals, obtains a centroid position of each cross-sectional line, obtains a centroid axis based on an average of the centroid positions, and sets each centroid position in a predetermined angle direction around the centroid position A section line point is set as a set section point, a longitudinal section including the set section point and the center of gravity axis is set, and a body outline line is created by connecting the set section points in the longitudinal section, and the body outline The present invention relates to a three-dimensional measurement system having a second determination unit that determines the position of a predetermined part based on a line.

  According to the present invention, the second determination unit creates a plurality of body outline lines by setting a plurality of cross-sectional point coordinate acquisition units that acquire the three-dimensional coordinates of the set cross-sectional points and a plurality of the vertical cross-sections in different directions. And calculating a horizontal distance between the set cross-sectional point on the body contour line and the centroid axis for each body contour line and creating a plurality of graphs, and the measurement based on the created graph The present invention relates to a three-dimensional measurement system including a part specifying unit that determines a detailed part of an object.

  Further, the present invention relates to a three-dimensional measurement system in which the part specifying unit specifies a dimension cross-sectional line of a predetermined part based on one graph selected from the plurality of graphs.

  Further, the present invention relates to a three-dimensional measurement system in which the part specifying unit specifies a dimensional cross-sectional line of another part from the cross-sectional line group based on the position of the specified part.

  In the present invention, the second determination unit further includes an automatic measurement processing unit, and the automatic measurement processing unit relates to a three-dimensional measurement system that measures the circumference of the specified dimension cross-section line as dimension data. .

  The present invention further includes a display unit that displays a table showing a list of site names created by the arithmetic and control unit, and an operation unit that can specify a desired site from the table displayed on the display unit. In addition, the present invention relates to a three-dimensional measurement system in which dimensional data of the part is displayed on the display unit by designating a desired part by the operation unit.

  The present invention further includes a display unit for displaying a three-dimensional image created by the arithmetic and control unit, and an operation unit capable of designating a desired part of the three-dimensional image. By specifying a desired part of the image, the dimension data of the part is related to the three-dimensional measurement system displayed on the display unit.

  The present invention also includes a step of taking a stereo image of a measurement object with a plurality of imaging units having two cameras, extracting feature points in a plurality of images obtained by the plurality of imaging units, and using the feature points as a basis. Creating a TIN of the object to be measured, setting a horizontal cross section at a predetermined interval in the vertical direction to the TIN, and obtaining a cross-section point that is an intersection of the horizontal cross section and the triangle constituting the TIN; Connecting the cross-section points in the horizontal cross-section to create a cross-section line; laminating the cross-section lines at predetermined intervals to create a cross-section line group; A step of creating, a step of acquiring each cross-sectional line point in a predetermined angle direction around the center of gravity position as a set cross-section point, setting a vertical cross-section including the set cross-section point and the barycentric axis, and Connect points in the longitudinal section A step of creating integrated contoured line between, those of the three-dimensional measurement method and a step of identifying a position of a predetermined region based on said body type contour line.

  Furthermore, the present invention provides a three-dimensional measurement system having a plurality of imaging units having two cameras and an arithmetic control device for controlling the imaging units. Stereo image, extract feature points in a plurality of images obtained by the plurality of imaging units, create a TIN of the measurement object based on the feature points, and cause the TIN to make vertical intervals at predetermined intervals. A horizontal section is set, a cross-section point that is an intersection of the horizontal cross-section and the triangle that constitutes the TIN is obtained, a cross-section line is created by connecting the cross-section points within the horizontal cross-section, and the cross-section line is defined. Stacked at intervals to create a group of cross-sectional lines, obtain the center of gravity position of each cross-sectional line, create a barycentric axis, and obtain the point of each cross-sectional line in a predetermined angle direction around the center of gravity position as a set cross-sectional point, The installation A vertical cross section including a cross-sectional point and the center of gravity axis is set, and a body outline line is created by connecting the set cross-section points in the vertical cross section, and a position of a predetermined part is specified based on the body outline line It relates to a three-dimensional measurement program.

  According to the present invention, a feature point is extracted from a plurality of image pickup units that have two cameras and perform stereo shooting of an object to be measured, and two images obtained by the image pickup unit, and common feature points on the image. A three-dimensional measurement system having a calculation control device for measuring a three-dimensional shape of the measurement object for each imaging unit and creating a three-dimensional image of the measurement object for each imaging unit. The plurality of imaging units are arranged so as to image the measurement object from a plurality of directions, and the arithmetic and control unit creates the three-dimensional shape data of the measurement object based on the TIN created based on the feature points. Then, a horizontal section is set at a predetermined interval in the vertical direction in the three-dimensional shape data, and a section line is created by connecting a section point that is an intersection of the horizontal section and the triangle constituting the TIN. A first determination unit that creates a group of cross-sectional lines by laminating at a predetermined interval, obtains a centroid position of each cross-sectional line and obtains a centroid axis based on an average of the centroid positions, A point of each cross-section line is set as a set cross-section point, a vertical cross-section including the set cross-section point and the center of gravity axis is set, and a body outline line is created by connecting the set cross-section points in the vertical cross-section. Since it has the 2nd judgment part which judges the position of a predetermined part based on an outline line, while specifying a desired part easily, measurement data can be easily used in a wide range according to a use. .

  Also, according to the present invention, the second determination unit sets a plurality of the cross-sectional point coordinate acquisition unit that acquires the three-dimensional coordinates of the set cross-sectional point and a plurality of the vertical cross-sections in different directions, so that the body outline line is Based on the created graph, a distance calculation unit that creates a plurality of graphs by creating and totaling the horizontal distance between the set cross-sectional point on the body contour line and the center of gravity axis for each body contour line Since it comprises the site | part specific part which determines the detailed site | part of the said measuring object, the said graph used according to the shape of a detailed site | part can be changed, and the erroneous detection of a detailed site | part can be prevented.

  According to the invention, the part specifying unit specifies the dimension cross-sectional line of the predetermined part based on one graph selected from the plurality of graphs, so that the position of the predetermined part is specified accurately. Can do.

  According to the invention, the part specifying unit specifies the dimension cross-sectional line of the other part from the cross-sectional line group based on the position of the specified part. Can be identified.

  According to the present invention, the second determination unit further includes an automatic measurement processing unit, and the automatic measurement processing unit measures the circumference of the specified dimension cross-section line as dimension data. Dimension data can be acquired and work labor can be reduced.

  According to the invention, there is further provided a display unit that displays a table showing a list of site names created by the arithmetic and control unit, and an operation unit that can specify a desired site from the table displayed on the display unit. In addition, when the desired part is specified by the operation unit, the dimension data of the part is displayed on the display unit, so that the dimension data of each part can be easily obtained by a simple operation.

  According to the present invention, it further comprises a display unit for displaying the three-dimensional image created by the arithmetic and control unit, and an operation unit capable of designating a desired part of the three-dimensional image. By designating a desired part of the three-dimensional image, the dimension data of the part is displayed on the display unit, so that the dimension data of each part can be easily obtained by a simple operation, and each part is assigned to the tertiary It can be easily recognized from the original image, and the operability can be improved.

  According to the present invention, a step of taking a stereo image of a measurement object with a plurality of imaging units having two cameras, and extracting feature points in a plurality of images obtained by the plurality of imaging units, the feature points A step of creating a TIN of the object to be measured based on the step, a step of setting a horizontal cross section at a predetermined interval in the vertical direction in the TIN, and obtaining a cross-section point that is an intersection of the horizontal cross section and a triangle constituting the TIN A step of connecting the cross-section points in the horizontal cross-section to create a cross-section line, a step of laminating the cross-section lines at a predetermined interval to create a cross-section line group, and determining the position of the center of gravity of each cross-section line A step of creating an axis, a step of obtaining each cross-sectional line point in a predetermined angle direction around the center of gravity position as a set cross-sectional point, and setting a vertical cross section including the set cross-sectional point and the barycentric axis, Connect the set cross-section points in the longitudinal section To create a body contour line and a step of identifying the position of a predetermined part based on the body contour line, so that a desired part can be easily identified and a wide range depending on the application. Measurement data can be used easily.

  Furthermore, according to the present invention, in a three-dimensional measurement system having a plurality of imaging units having two cameras and an arithmetic control device for controlling the imaging units, the arithmetic control device is measured by the plurality of imaging units. Stereo imaging of the object, feature points in a plurality of images obtained by the plurality of imaging units are extracted, a TIN of the measurement object is created based on the feature points, and the TIN has a predetermined vertical direction A horizontal cross section is set at intervals, a cross-section point that is an intersection of the horizontal cross-section and the triangle that constitutes the TIN is obtained, a cross-section line is created by connecting the cross-section points within the horizontal cross-section, and the cross-section line Are created at a predetermined interval to create a group of cross-sectional lines, determine the position of the center of gravity of each cross-sectional line, create the center of gravity, and obtain the point of each cross-sectional line in the predetermined angle direction around the position of the center of gravity as the set cross-sectional point Let A vertical cross section including the set cross-sectional point and the center of gravity axis is set, and a body outline line is created by connecting the set cross-section points in the vertical cross section, and a position of a predetermined part is determined based on the body outline line Since it is specified, it is possible to easily specify a desired part and to exhibit an excellent effect that measurement data can be easily used in a wide range according to the application.

It is a block diagram of the three-dimensional measurement system which concerns on the Example of this invention. It is a block diagram of the arithmetic and control unit in the three-dimensional measurement system of the present invention. It is explanatory drawing which shows the state which made the binary image from the front direction the image which imaged the three-dimensional shape data represented by TIN by the parts recognition process part. It is explanatory drawing which shows the binary image in which the 1st grouping process was performed by the parts recognition process part. It is explanatory drawing which shows the binary image by which the 1st separation process was performed based on the detected boundary by the parts recognition process part. It is explanatory drawing which shows the binary image by which the 2nd separation process was performed based on the inseam position detected by the parts recognition process part. It is explanatory drawing which shows the binary image by which the 2nd grouping process was performed by the parts recognition process part. It is explanatory drawing which shows the binary image by which each part was recognized by the parts recognition process part, and was grouped for every part. It is explanatory drawing explaining the detection of the boundary of a measuring object by a parts recognition process part, (A) shows the grid group of a horizontal line, (B) shows the plane cross section of a black grid group, (C) is The graph of the number of counts when the distance between the vertex of the triangle and the dividing line is within the threshold is 1 count is shown. It is explanatory drawing explaining the detection of the crotch position of a measurement object by a parts recognition process part, (A) shows the grid group of a horizontal line, (B) shows the plane cross section of a black grid group, (C) Shows a graph representing the distance between the apex of the triangle and the dividing line. It is explanatory drawing explaining the state by which the crotch position of the measuring object was detected by the parts recognition process part, (A) shows the grid group of a horizontal line, (B) shows the plane cross section of a black grid group, (C) shows a graph representing the distance between the apex of the triangle and the dividing line. It is explanatory drawing explaining another method for detecting the crotch position of a measuring object by a parts recognition process part, (A) shows a grid group of a horizontal line, (B) is a plane cross section of a black grid group (C) shows a graph representing the distance between the apex of the triangle and the dividing line. It is a top view which shows the cross-sectional line produced by the cross-section line creation part. It is a graph which shows the section line by which cylindrical coordinates were changed by the 3rd noise processing part. It is a perspective view which shows the trunk | drum sectional line group by which the sectional line of the trunk | drum site | part was laminated | stacked. It is a front view which shows TIN of the trunk | drum site | part interpolated. It is a flowchart explaining the detailed site | part determination processing based on the Example of this invention. It is explanatory drawing explaining the relationship between a fuselage cross-sectional line group, a gravity center axis, and a setting cross-section point, (A) shows the front view of the said fuselage cross-sectional line group, (B) shows the top view of this fuselage cross-sectional line group ing. It is the graph which totaled the distance between a gravity center axis | shaft and a setting cross-section point in case a setting cross-section point is located in the front side diagonal direction of a trunk | drum cross-section line group. It is the graph which totaled the distance between a gravity center axis | shaft and a setting cross-section point when a set cross-section point is located in the back side diagonal direction of a trunk | drum cross-section line group. It is the graph which totaled the distance between a gravity center axis | shaft and a setting cross-section point when a setting cross-section point is located in the side surface direction of a fuselage cross-section line group. It is the graph which totaled the distance between a centroid axis and a set section point when a set section point is located in the front direction of a body section line group.

  Embodiments of the present invention will be described below with reference to the drawings.

  First, referring to FIG. 1, a three-dimensional measurement system 1 according to an embodiment of the present invention will be described.

  In FIG. 1, reference numeral 2 denotes a measurement object such as a human body, and reference numerals 3 to 10 denote cameras that are image pickup units having an image pickup element such as a CCD or CMOS sensor, which is an aggregate of a large number of pixels. Indicates a feature point mark irradiation unit such as a projector.

  The cameras 3 to 10 have two cameras 3 and 4, cameras 5 and 6, cameras 7 and 8, and cameras 9 and 10 on four sides of the measurement object 2 so as to surround the measurement object 2. One set is arranged in the same horizontal plane at a predetermined interval, and each set of the cameras 3, 4, the cameras 5, 6, the cameras 7, 8, and the cameras 9, 10 is arranged in the circumferential direction, etc. Arranged at angular intervals.

  The distance between the optical axes of the cameras 3 and 4, the distance between the optical axes of the cameras 5 and 6, the distance between the optical axes of the cameras 7 and 8, and the distance between the optical axes of the 9 and 10 are respectively Equal and known values.

  The feature point mark irradiators 12 to 15 are disposed between the cameras 3 and 4, the cameras 5 and 6, the cameras 7 and 8, and the cameras 9 and 10, respectively. The optical axis of the unit 12 is the center between the optical axes of the cameras 3 and 4, the optical axis of the feature point mark irradiation unit 13 is the center between the optical axes of the cameras 5 and 6, and the light of the feature point mark irradiation unit 14. The axis is set to be the center between the optical axes of the cameras 7 and 8, and the optical axis of the feature point mark irradiation unit 15 is set to be the center between the optical axes of the cameras 9 and 10.

  Further, the cameras 3 and 4 and the feature point mark irradiation unit 12, the cameras 5 and 6 and the feature point mark irradiation unit 13, the cameras 7 and 8 and the feature point mark irradiation unit 14, the cameras 9 and 10 and The feature point mark irradiating section 15 is set so that the optical axes are parallel to each other, and the camera 3 and 4, the cameras 5 and 6, the cameras 7 and 8, and the cameras 9 and 10 are used to The measurement object 2 can be photographed in stereo. The optical axis is not parallel and may be directed inward. In this case, the angles of the optical axes between the cameras 3 and 4, between the cameras 5 and 6, between the cameras 7 and 8, and between the cameras 9 and 10 are also known.

  The feature point mark irradiation units 12 to 15 irradiate the measurement object 2 with white light or pattern light, and form feature points on the measurement object 2 by the irradiated white light or pattern light.

  An imaging unit is configured by the cameras 3 to 10 arranged side by side and the feature point mark irradiation units 12 to 15 arranged between the cameras 3 to 10, respectively. , 4 and the feature point mark irradiation unit 12 constitute a first imaging unit 16, and the cameras 5 and 6 and the feature point mark irradiation unit 13 constitute a second imaging unit 17, and the cameras 7, 8, and A third imaging unit 18 is configured by the feature point mark irradiation unit 14, and a fourth imaging unit 19 is configured by the cameras 9 and 10 and the feature point mark irradiation unit 15. In FIG. 1, the cameras of each set of the imaging units 16 to 19 are arranged in the horizontal direction, but needless to say, they may be arranged in the vertical direction.

  The cameras 3 to 10 are connected to an input / output control unit 21 such as a switching hub by communication means such as a LAN, and the images captured by the imaging units 16 to 19 are linked to the captured camera in the input state. The data is input to the arithmetic control device 22 such as a PC via the output control unit 21.

  The arithmetic and control unit 22 is electrically connected to an operation unit 23 such as a keyboard and a mouse and a display unit 24 such as a monitor. When a desired command is input to the operation unit 23, predetermined processing is performed on the image input to the arithmetic and control unit 22, and the captured image, the processed image, or the measurement result is displayed on the display unit. 24 is displayed. Note that the display unit 24 may be used as the operation unit 23 by using the display unit 24 as a touch panel.

  Next, the arithmetic control unit 22 will be described with reference to FIG. In the following description, the case where the measurement object 2 is a human body will be described.

  The calculation control device 22 includes a calculation control unit 25 such as a CPU, a first determination unit 26 that determines an approximate part of the measurement object 2, and a second determination unit 27 that determines a detailed part of the measurement object 2. And a storage unit 28 such as a memory or HDD.

  The arithmetic control unit 25 activates and develops a program to be described later, and executes predetermined processing based on the program.

  The first determination unit 26 includes a matching processing unit 29, a first noise deletion processing unit 31, a parts recognition processing unit 32, a second noise deletion processing unit 33, a cross-section line creation unit 34, and a third noise deletion. A processing unit 35, a cross-sectional line interpolation unit 36, and a three-dimensional shape interpolation unit 37 are included.

  The matching processing unit 29 creates three-dimensional point cloud data in which each point has coordinates (Xi, Yi, Zi) based on each of the two images acquired by the imaging units 16 to 19. A three-dimensional point network data is connected by a line, and a TIN (Triangulated Irregular Network) that expresses the three-dimensional shape of the measurement object 2 by a set of triangles having different sizes is created. Thereby, three-dimensional shape data expressed by the TIN is created.

  Based on the created TIN, the first noise removal processing unit 31 performs the measurement of the three-dimensional shape data of the measurement object 2 in a total of eight directions including the front direction, the back direction, the both side surfaces direction, and the oblique surface direction. A total of eight first binary images, which are silhouette images, are created, predetermined processing such as expansion / contraction processing is performed on the first binary image, and noise is deleted from the TIN. In the above description, the number of directions is eight, but the number of directions may be changed such as six directions and ten directions.

  The parts recognition processing unit 32 creates a second binary image based on the front direction with respect to the TIN from which noise has been deleted by the first noise deletion processing unit 31. And has a function of recognizing a total of five parts of the legs, arms, and trunk in the three-dimensional shape data expressed by the TIN and grouping the points constituting the TIN.

  The second noise deletion processing unit 33 deletes noise from the TIN based on the part information recognized by the parts recognition processing unit 32.

  For the TIN from which the noise has been deleted by the second noise deletion processing unit 33, the cross-section line creation unit 34 performs contour lines at predetermined intervals in the vertical direction with respect to the three-dimensional shape data represented by the TIN. pull. The points where the contour lines intersect with the triangles constituting the TIN (cross-sectional points), that is, the intersections of the horizontal cross-sections set at a predetermined interval and the triangles, are found on the horizontal cross-section, By connecting with a line, a cross-sectional line that is an outline of the horizontal cross section of the TIN is obtained. In addition, the cross-sectional line creation unit 34 stacks the acquired cross-sectional lines for each grouped part, and stores them in the storage unit 28 as a cross-sectional line group for each part.

  The third noise removal processing unit 35 performs cross-sectional line analysis for converting the cross-sectional line into a cylindrical coordinate system on the acquired cross-sectional line, and noise from the TIN based on the cross-sectional line information of the converted cylindrical coordinate system. Delete.

  The cross-sectional line interpolation unit 36 re-converts the cross-sectional line converted into the cylindrical coordinate system into a normal orthogonal coordinate system, and the first noise deletion processing unit 31, the second noise deletion processing unit 33, and the third By performing a linear interpolation process on the cross-sectional line in which the missing portion is generated by the noise removing process in the noise removing processing unit 35, a virtual interpolation point is created in the missing portion and the cross-sectional line is interpolated.

  The three-dimensional shape interpolation unit 37 creates a triangle that fills the missing portion generated in the TIN by using the interpolation point and the sectional line created by the section line interpolation unit 36, and the TIN having no missing portion and the TIN. It has a function to create three-dimensional shape data expressed by

  The second determination unit 27 includes a center-of-gravity axis creation unit 38, a cross-section point coordinate acquisition unit 39, a distance calculation unit 40, a part specifying unit 41, an automatic measurement processing unit 44, and a blindfold processing unit 45. Have. In this embodiment, the part specifying unit 41 further includes a body part specifying unit 42 as a first part specifying unit and a face part specifying unit 43 as a second part specifying unit.

  The center-of-gravity axis creation unit 38 designates a body cross-sectional line group 69 described later from the cross-sectional line group created by the cross-sectional line creation unit 34 and interpolated by the cross-sectional line interpolation unit 36, and each cross-sectional line 64 The center of gravity position (center of gravity X) is obtained for each, the center of gravity position is averaged to obtain the average center of gravity position, and a vertical line passing through the average center of gravity position is created as a center of gravity axis 75 (described later).

  The cross-sectional point coordinate acquisition unit 39 sets a plurality of angles θ in advance around the center of gravity X of the cross-sectional line 64, rotates a reference line 79 (see FIG. 18B) for each set angle, The intersection of the reference line 79 and the cross-sectional line 64 is acquired as a set cross-sectional point 76 (described later). The set cross-section points 76 are acquired in the same manner for all cross-section lines 64, and the three-dimensional coordinates of each set cross-section point 76 are acquired.

  The distance calculation unit 40 sets a longitudinal section 77 (described later) that is a plane including each set section point 76 and the center of gravity axis 75 at the same set angle, and the set section existing on the longitudinal section 77. The series of points 76 forms a body outline line 78 (see FIG. 18A) described later. The distance calculation unit 40 calculates a horizontal distance between the center of gravity axis 75 and the set cross-sectional point 76 based on the three-dimensional coordinates of the set cross-sectional points 76 on the vertical cross section 77 and calculates A horizontal distance is totaled for each cross-sectional line 64 to create a graph (described later).

  The part specifying unit 41 specifies the dimension cross-sectional line required for each part based on the graph created by the distance calculating unit 40. For example, when obtaining a dimensional cross-section line for a body part, the body part specifying unit 42 operates, and the body part specifying unit 42 selects a graph indicating a characteristic shape of the body part and displays the selected graph on the selected graph. Based on the body part, the dimensional cross section lines of the neck, bust, waist, and hip are specified, and each dimensional cross section line is specified as the neck part, bust part, hip part, and waist part. In FIG. 18B, the lower side with respect to the paper surface is the front side, and when the neck part and the bust part are specified, the set cross-sectional point 76 exists in the front side oblique direction of the cross-sectional line 64. In this case, for example, a graph in which the cases of −135 ° or −45 ° in FIG. 18B are tabulated is used. That is, a graph is used in which the set cross-sectional points 76 existing on the vertical cross-section 77 set in the diagonal direction on the front side are tabulated. Further, when specifying the hip region, a graph in which the set cross-sectional points 76 are aggregated when the cross-sectional line 64 is present in an oblique direction on the back side, for example, 50 ° or 130 ° in FIG. 18B is used. When the waist portion is specified, a graph in which the set cross-sectional points 76 are tabulated in the side surface direction of the cross-sectional line 64, for example, 180 ° or 0 ° in FIG. 18B is used.

  Next, when the dimension cross-section line is specified for the face part, the face part specifying unit 43 is operated, the face part specifying unit 43 selects a graph effective for specifying the face part, and the graph From the position of the neck part specified by the body part specifying unit 42, the dimensional cross section line of the nose part is specified. Further, the dimension cross-sectional line of the eye part is specified based on the dimension cross-sectional line of the nose part, and each dimension cross-sectional line is specified as the nose part and the eye part. When specifying the nasal part and the eye part, a graph in which the set cross-sectional point 76 is tabulated in the front direction of the cross-sectional line 64, for example, −90 ° in FIG. 18B is used. It is done.

  The automatic measurement processing unit 44 acquires the perimeter of each dimensional cross-sectional line specified by the body part specifying unit 42 and the face part specifying unit 43 as dimension data of each specific part.

  The blindfold processing unit 45 paints a texture image of a portion corresponding to the eye part specified by the face part specifying unit 43 with a single color.

  The storage unit 28 includes a program for driving and controlling the imaging units 16 to 19, matching based on images captured by the imaging units 16 to 19, and the three-dimensional shape data represented by the TIN and the TIN. , A program for deleting noise from the TIN based on the first binary image, various parts of the measurement object 2 are recognized based on the second binary image, and part information A program for grouping the TIN based on the above, a program for removing noise from the TIN based on the part information, and a contour line across the triangle constituting the TIN are drawn to obtain the cross section line 64 of the TIN. Program, a program for deleting the noise of the TIN based on the acquired cross-sectional line 64, and the cross-sectional line 6 generated by deleting various noises. A program for creating a virtual interpolation point in a blank part of the image, a triangle is created based on the interpolation point and the section line 64, and the TIN having no blank and the three-dimensional shape data represented by the TIN are created. And a program such as a first determination program for overall control of the first determination unit 26 are stored.

  Further, the storage unit 28 obtains the center of gravity X of each cross-sectional line 64 interpolated with a blank portion, and creates a center of gravity axis 75 based on the obtained center of gravity X, a preset setting direction. A program for acquiring the set cross-section point 76 based on an angle, and for acquiring the three-dimensional coordinates of the set cross-section point 76, for setting the longitudinal section 77 including the center of gravity axis 75 and the set cross-section point 76. A program, a program for calculating the distance between the center of gravity axis 75 and the set section point 76 for each section line 64 based on the acquired three-dimensional coordinates of the set section point 76, and a tabulated center of gravity axis 75 And a program for specifying a cross section line 64 corresponding to a neck part, a bust part, a hip part, and a waist part that are set in advance from the body part based on the distance of the set cross-sectional point 76, A program for specifying the cross-sectional line 64 corresponding to the nose part and the eye part from the face part based on the neck part specified from the above, and for obtaining the perimeter of the cross-sectional line 64 corresponding to the specified part Programs such as a program, a program for painting a texture image of a portion corresponding to the specified eye part with a single color, and a second determination program for controlling the second determination unit 27 in an integrated manner are stored.

  Further, the storage unit 28 stores captured image data, image data that has been subjected to image processing, setting values and threshold values for performing various types of processing, and data created by various types of processing. Yes.

  The matching processing unit 29 includes a program for creating matching, the TIN, and three-dimensional shape data expressed by the TIN, and the arithmetic control unit 25. The first noise deletion processing unit 31 includes a program for deleting noise from the TIN based on the first binary image and the arithmetic control unit 25. The parts recognition processing unit 32 recognizes various parts of the measurement object 2 based on the second binary image and groups the TIN based on part information and the arithmetic control unit 25. It consists of. The second noise deletion processing unit 33 includes a program for deleting noise from the TIN based on part information and the arithmetic control unit 25. The section line creation unit 34 is composed of a program for obtaining a section line 64 of the TIN by drawing contour lines crossing the triangles constituting the TIN and the arithmetic control unit 25. The third noise deletion processing unit 35 includes a program for deleting the TIN noise based on the acquired sectional line 64 and the arithmetic control unit 25. The cross-sectional line interpolation unit 36 includes a program for creating a virtual interpolation point in a blank portion of a cross-sectional line generated by deleting various noises and the arithmetic control unit 25. The three-dimensional shape interpolating unit 37 creates a triangle based on the interpolation point and the cross-sectional line 64 to create the TIN having no space and the three-dimensional shape data represented by the TIN and the calculation. And a control unit 25.

  The center-of-gravity axis creation unit 38 includes a program for creating the center-of-gravity axis 75 based on the center of gravity X of each sectional line 64 and the arithmetic control unit 25. The section point coordinate acquisition unit 39 includes a program for acquiring the three-dimensional coordinates of the set section point 76 of the set direction angle and the calculation control unit 25. The distance calculating unit 40 is a program for setting the longitudinal section 77, a program for calculating and summing the distance between the set section point 76 acquired for each section line 64 and the barycentric axis 75, and the calculation control section. 25. Further, the torso part specifying unit 42 specifies a cross section line 64 corresponding to the neck part, the bust part, the hip part, and the waist part from the torso part based on the total distance between the center of gravity X and the set cross section point 76. The calculation control unit 25 is configured. The face part specifying unit 43 includes a program for specifying the cross-sectional line 64 corresponding to the nose part and the eye part based on the specified neck part and the arithmetic control unit 25. The automatic measurement processing unit 44 includes a program for obtaining the perimeter of the section line 64 corresponding to the specified site and the arithmetic control unit 25. The blindfold processing unit 45 includes a program for painting a texture image of a portion corresponding to the specified eye part with a single color and the arithmetic control unit 25.

  Next, with regard to the three-dimensional measurement process using the three-dimensional measurement system 1, first, an outline part specifying process by the first determination unit 26 will be described.

  First, the feature point mark irradiation unit 12 of the first imaging unit 16 can irradiate two types of light, white light and pattern light, and the first imaging unit 16 applies white light to the measurement object 2. Two types of images, that is, a state in which light is irradiated and a state in which pattern light is irradiated on the measurement object 2 are acquired two by two. An image captured by the first imaging unit 16 is input to the arithmetic control device 22 via the input / output control unit 21.

  When the captured image is input to the arithmetic and control unit 22, feature points are first extracted from the image irradiated with pattern light by edge detection or the like by the matching processing unit 29, and at least three of the extracted feature points are extracted. A point is selected, and the two feature images are matched by associating the selected feature points with each other, and the three-dimensional coordinates of the point on the image are calculated.

  For image matching, an SSDA method (Sequential Similarity Detection Algorithm), a normalized cross-correlation method, a least square matching method, or the like is used.

  Thus, the three-dimensional coordinates of the feature points are obtained by photogrammetry based on the two images based on the position of the first imaging unit 16 that is already known, and adjacent feature points are defined with the feature points as the vertices of a triangle. By creating a TIN by connecting with a line, three-dimensional shape data represented by the TIN of the measurement object 2 is created. Then, an image irradiated with white light is pasted on the TIN (texture mapping).

  The second imaging unit 17, the third imaging unit 18, and the fourth imaging unit 19 also create a TIN in the same manner as the first imaging unit 16, and a tertiary expressed in TIN covering almost all directions. Original shape data can be created.

  When the TIN is created by the matching processing unit 29, the first noise deletion processing by the first noise deletion processing unit 31 is performed next. First, the first noise removal processing unit 31 uses a three-dimensional shape data expressed by the TIN with reference to the front direction as a grid of a predetermined size, for example, a square grid of 5 mm × 5 mm in actual size. When grids are arranged on a grid sheet arranged in a matrix and there are at least one triangle vertex in each grid, the grid is white, and when there are no triangle vertices in the grid Using the grid as black, the first binary image in which the silhouette image of the three-dimensional shape data expressed by the TIN is displayed in white is created.

  After the creation of the first binary image, the first noise deletion processing unit 31 performs a contraction process on the first binary image, and then repeats the expansion process three times. By performing the contraction process, it is possible to delete the isolated noise at a position away from the silhouette image in the first binary image, and by repeating the expansion process three times, Missing can be prevented.

  In this embodiment, since the density of the vertices of the triangle is not constant, the first binary image is subjected to three expansion processes for one contraction process. It goes without saying that the number of expansion processes is appropriately selected according to various conditions such as the size of one grid.

  After the contraction / expansion processing, the first noise deletion processing unit 31 deletes the vertices of the triangles existing in the black grid as noise.

  In the above, the first binary image is created based on the three-dimensional shape data expressed by the TIN with respect to the front direction. However, the back direction, both side directions, and the oblique direction of the three-dimensional shape data. The first binary image is also created for the seven directions, and noise is deleted in the same manner as described above.

  Therefore, in the first binary image of the three-dimensional shape data expressed by the TIN with reference to a total of eight directions including the front direction, the back direction, the both side directions, and the oblique direction, a part of the first and second images are displayed. In the value image, even if it is a vertex of a triangle existing in the white grid, if there is another first binary image in which the vertex of the triangle exists in the black grid, the vertex of the triangle is deleted as noise. The The same applies to the case where the first binary image is created from different directions such as six directions and ten directions.

  When the noise of the TIN is deleted by the first noise deletion processing unit 31, the parts recognition processing by the parts recognition processing unit 32 is performed next.

  The parts recognition processing unit 32 first converts the three-dimensional shape data expressed by the TIN with the front direction as a reference, for example, a grid sheet in which a large number of grids are arranged in a matrix with a 20 mm × 20 mm square grid as one grid. If there is even one triangle vertex in each grid as shown in Fig. 3, the grid is black, and if there is no triangle vertex in the grid, the grid is A second binary image 47 having a white grid is created.

  In FIG. 3 and FIGS. 4 to 8 to be described later, different groups are displayed in different colors (in the drawing, the difference in color is represented by a difference in brightness) for convenience.

  Subsequently, the parts recognition processing unit 32 searches the grid groups of the horizontal lines in the second binary image 47 one by one in order from the left, and sets the portions where the black grids are continuous as one group. A first grouping process for grouping is performed. For example, as shown in FIG. 4, the first group 48 is defined from the grid where black is first detected from the left to the grid where white is detected right next to the right side of the first group 48. From the grid in which black is detected to the grid in which white is detected on the right side is defined as the second group 49, and on the right side of the second group 49, white is detected on the right side from the grid in which black is detected. Grouping is performed as a third group 50 up to the grid.

  In FIG. 9, after the first grouping process, the distance between the vertices of the triangles existing in one grid in the grid group of the horizontal lines of the second binary image 47 (see FIG. 9A) is If the calculated distance is within a preset threshold, it is counted as one count, and the count is graphed as shown in FIG. 9C. Further, the grid having the largest count number, that is, the largest number of triangle vertices existing in one grid is detected as the boundary 52 of the measurement object 2. In FIG. 9, 53 indicates a grid group of one horizontal line extracted from the second binary image 47 of FIG. 4, and 54 a and 54 b of FIG. 9B indicate the plane of the measurement object 2. It is a cross section, for example, shows a leg part of a human body, and corresponds to the first group 48 and the second group 49 in FIG.

  After the detection of the boundary 52, the first to third groups 48 to 50 grouped based on the detected boundary 52 are, for example, grids on which the boundary 52 is detected as shown in FIG. A first separation process to be separated is performed, and positions such as crotch and heel are recognized.

  However, since the detection of the boundary 52 described above is not sufficient for accurate detection of the crotch position because of the clothes worn, etc., the parts recognition processing unit 32 then determines the crotch position (branch position). Detection is performed.

  When detecting the crotch position, first, of the vertical lines in the second binary image 47, a range set in advance as the position relating to the torso, for example, 25% to 65% from the lower end of the TIN. 10, for example, as shown in FIG. 10, the flat sections 54 a and 54 b are equally divided into a front side and a back side by a dividing line 55 passing through the center of the figure of the flat sections 54 a and 54 b. At the same time, only the data of the three-dimensional shape data expressed by the TIN on the divided front side (lower side in FIG. 10B) is extracted.

  After extracting the data of the three-dimensional shape data on the front side, the shortest distance from the apex of the triangle existing in front of the flat sections 54a and 54b to the dividing line 55, that is, the length of the perpendicular line is calculated. The distance is graphed (see FIG. 10C). The flat cross sections 54a and 54b of the created graph have peak portions 56a and 56b formed by increasing the distance between the dividing line 55 and the apex of the triangle, and the dividing line 55 is formed between the flat cross sections 54a and 54b. The trough 57 is formed by shortening the distance between the apexes of the triangle. As shown in FIG. 11, the horizontal line grid group 53 is sequentially searched one line upward from the preset lower end, that is, 25% of the horizontal line grid group 53 from the lower end of the TIN. The grid group 53 of the horizontal line in which the peak portion 56 and the valley portion 57 are no longer detected is detected as a crotch position.

  In the above-described inseam position detection process, in order to detect an accurate inseam position, the measurement object 2, that is, the person to be measured needs to be standing at an accurate measurement position. It is difficult to always stand at an accurate measurement position. Therefore, in this embodiment, the part recognition processing unit 32 performs correction processing on the measurement position.

  Specifically, by detecting the inclination angle of the common tangent line in contact with the flat sections 54a and 54b, it is possible to detect a deviation from the accurate measurement position, and to rotate the X axis and the Y axis by the inclination angle of the common tangent line. Thus, the measurement position can be corrected.

  In the above description, the crotch position is detected by detecting the peak portions 56a and 56b and the valley portion 57 in the graph. However, as shown in FIG. Focusing on the portion of the distance of 85% with respect to the distance to the apex of the portion 56, whether or not there are two flat plateau portions 58a and 58b formed on the flat cross sections 54a and 54b in the graph. The position may be detected. The distance for detecting the plateau portions 58a and 58b is not necessarily 85% of the distance to the apex of the mountain portion 56, and is appropriately selected according to conditions.

  When the crotch position in the second binary image 47 is detected, based on the detected crotch position, that is, the boundary between both legs, a group of the second binary images 47 is displayed as shown in FIG. Further, a second separation process for separation is performed. By performing the second separation process, the boundary between the first group 48 and the second group 49 is further clarified.

  Next, a second grouping process is performed by the parts recognition processing unit 32 based on the second binary image 47 grouped by the second separation process.

  In the second grouping process, as shown in FIG. 7, black grid groups of horizontal lines adjacent in the vertical direction are connected or separated, and the first group 48, the second group 49, and the third group 50. The fourth group 59, the fifth group 60, the sixth group 61, and the seventh group 62 are grouped into seven groups.

  Specifically, the horizontal deviation between the horizontal center of the black grid group of the horizontal line and the horizontal center of the black grid group of the horizontal line adjacent to the vertical direction of the black grid group (pixels in the pixel) If the calculated distance is within a preset threshold, the black grid groups are connected as the same group, and if the calculated distance is greater than the threshold, the black grid groups Are separated as a separate group.

  If the black grid group of another group is connected to the black grid group of the horizontal line and there is no black grid group that can be connected to the upper and lower sides of the black grid group of one group, The above processing is performed after regarding the black grid group as a black grid group of one group.

  Finally, the parts recognition processing unit 32 automatically recognizes the second binary image 47 grouped by the second grouping process in order from the lower group, such as both legs, torso, both arms, and the like. The fourth group 59 in FIG. 7 and the seventh group 62 existing on the fourth group 59 are connected, and as shown in FIG. 8, the five parts of both arms, the trunk, and both legs are recognized. The second binary image 47 which does not depict the other consisting of the five groups is generated.

  After the part recognition, the second noise deletion processing unit 33 performs a second noise deletion process. The second noise removal processing unit 33 acquires which part group each vertex of the triangle belongs to based on the part information grouped by the parts recognition processing unit 32. The triangles belonging to the group of parts are deleted as noise, and the triangles existing outside the five parts of both legs, torso, and both arms, for example, the third group 50 recognized as others in FIG. 7 are deleted as noise. .

  It should be noted that each part in FIG. 8 is set as a boundary line, and triangles that cross the boundary line are not deleted. For example, using the boundary between the first group 48 and the third group 50 as a boundary line, a triangle constituted by a point belonging to the first group 48 and two points belonging to the third group 50 is not deleted. Even if the triangle vertices belong to all different parts, they are not deleted in the same way when they cross the boundary line.

  After the second noise removal processing by the second noise removal processing unit 33, the cross-section line creation unit 34 draws contour lines at a predetermined interval with respect to the three-dimensional shape data represented by the TIN, and the contour line is the TIN. By obtaining a point (cross-sectional point) that intersects with the triangle that constitutes, a cross-sectional line 64 (described later) that is an outline of the flat cross-section 54 of the TIN is acquired.

  As shown in FIG. 13, the cross-sectional line 64 is obtained from the first partial cross-sectional line 65 acquired from the image captured by the first imaging unit 16 and the image captured by the second imaging unit 17. The acquired second partial cross-sectional line 66, the third partial cross-sectional line 67 acquired from the image captured by the third imaging unit 18, and the image acquired by the fourth imaging unit 19 The cross section line 64 includes four partial cross section lines 68, and the cross section lines 64 are acquired at intervals of, for example, 2.5 mm over the entire length in the vertical direction.

  When the cross-sectional line 64 is acquired, it is identified which part cross-sectional line each partial cross-sectional line belongs to, and which imaging unit the TIN of the cross-sectional line is the partial cross-sectional line. Information to do is acquired at the same time.

  Further, the acquired cross-sectional line 64 is stored in the storage unit 28 for each part as a cross-sectional line group in which the cross-sectional lines 64 are laminated as in the body cross-sectional line group 69 shown in FIG. 15, for example. Is done.

  When the cross-sectional line 64 of the TIN is created, the third noise deletion processing unit 35 performs a third noise deletion process.

  The third noise removal processing unit 35 first obtains the center of gravity of the figure formed by the cross-sectional line 64 and converts the cross-sectional line 64 into a cylindrical coordinate system. FIG. 14 is a graph created by connecting the cross-sectional points with straight lines, taking the cross-sectional line 64 converted into a cylindrical coordinate system by taking the distance r between the center of gravity and the cross-sectional point on the vertical axis and the inclination θ on the horizontal axis. It is. Further, as shown in FIGS. 13 and 14, the first to fourth partial cross-sectional lines 65 to 68 are partially overlapped (71 and 72 in FIGS. 13 and 14), and partly do not overlap. The sectional line 64 is configured in a separated state (73 and 74 in FIGS. 13 and 14).

  Preferably, the center of the figure is obtained by using data excluding cross-sectional points at both ends of the first to fourth partial cross-sectional lines 65 to 68 of the cross-sectional line 64 described later.

  When the section line 64 is converted into the cylindrical coordinate system, the noise section point is next detected. The third noise removal processing unit 35 first includes an overlapping portion where the partial cross-sectional lines overlap, in FIG. 13 and FIG. 14, the overlapping portion 71 of the second partial cross-sectional line 66 and the third partial cross-sectional line 67, and the About the overlapping part 72 of the third partial cross-sectional line 67 and the fourth partial cross-sectional line 68, at a predetermined number of cross-sectional points on the end side of each of the first to fourth partial cross-sectional lines 65 to 68, for example, at five cross-sectional points The inclination angle of the tangent line is calculated. If the inclination angle is equal to or larger than the set threshold value, the point is detected as a noise cross-section point, and the TIN triangle corresponding to the noise cross-section point is deleted as noise.

  Further, after detecting and deleting the noise cross-section points for the overlapping portions 71 and 72, the range where the two partial cross-section lines overlap is calculated as 100%, and the overlapping range (the matching range) ) Is a predetermined value, for example, 33%, a point outside the predetermined value range is detected as a noise cross-sectional point, and the TIN triangle corresponding to the noise cross-sectional point is deleted as noise.

  Next, in the separated portion where the partial cross-sectional lines do not overlap, in FIG. 13 and FIG. 14, the defect portion 73 of the first partial cross-sectional line 65 and the second partial cross-sectional line 66, the fourth partial cross-sectional line 68 and the first Also for the missing portion 74 of the first partial sectional line 65, a predetermined number of sectional points on the end side of each of the first to fourth partial sectional lines 65 to 68, for example, calculate the inclination angle of the tangent line at the five sectional points, If the inclination angle is equal to or greater than the set threshold value, the point is detected as a noise cross-section point, and the TIN triangle corresponding to the noise cross-section point is deleted as noise.

  When the noise is deleted by the first noise deletion process, the second noise deletion process, and the third noise deletion process, the cross-sectional line 64 may have a defective portion where the partial cross-sectional lines do not overlap each other. The cross-sectional line interpolation unit 36 re-converts the cross-sectional line 64 converted into the cylindrical coordinate system into a normal coordinate system, and creates virtual interpolation points in the missing portions 73 and 74 by linear interpolation processing. By connecting the points with a straight line, the missing portion of the cross-sectional line 64 is compensated.

  Finally, the three-dimensional shape interpolation unit 37 creates a new triangle from the section line 64 and the interpolation point for the missing portion of the TIN generated by each noise removal process. The three-dimensional shape interpolation unit 37 creates a triangle in the missing portion, thereby creating the TIN having no blank portion and the three-dimensional shape data represented by the TIN, which are displayed on the display unit 24 as shown in FIG. Such data is stored in the storage unit 28 for each part, such as a body part TIN70 or a three-dimensional image obtained by pasting an image of the body part TIN70 irradiated with white light.

  Next, the detailed part determination process of the measurement object 2 using the three-dimensional measurement system 1 will be described with reference to the flowchart shown in FIG. In performing the detailed part determination process, whether the body shape of the human body that is the measurement object 2 is thin or fat is input in advance. The body shape may be set by inputting height and weight as initial input items and obtaining an index such as BMI.

  STEP: 01 First, as shown in FIG. 18A, the center-of-gravity axis creation unit 38 requires a particularly detailed body part to be identified from the cross-sectional line groups of both arms, the body, and both legs. The body section line group 69 of the part is designated. Next, the center-of-gravity axis creation unit 38 obtains the center-of-gravity X of each cross-sectional line 64 constituting the body cross-sectional line group 69.

  (Step 02) After the center of gravity X of each cross-sectional line 64 is obtained, the center-of-gravity axis creation unit 38 obtains the average of the X-coordinate and the average of the Y-coordinate of each center-of-gravity X, respectively. ) And a vertical straight line passing through the average barycentric coordinates (Xm, Ym) is created as the barycentric axis 75 of the body section line group 69.

  STEP: 03 In parallel with STEP: 02, the cross-section point coordinate acquisition unit 39 rotates the reference line 79 around a center of gravity X of the cross-section line 64 in a plurality of preset directions (set direction angles), The intersection of the reference line 79 and the cross section line 64 is acquired as the set cross section point 76. The set cross-sectional points 76 are acquired for each cross-sectional line 64 for each set direction angle, and the three-dimensional coordinates of the acquired set cross-sectional points 76 are acquired. In FIG. 18B, only the set cross-sectional point 76 in the diagonal direction on the front side when the reference line 79 is rotated by the set direction angle θ is shown.

  (Step 04) When the barycentric axis 75 and the set cross-sectional point 76 are acquired, the distance calculation unit 40 is a plane including the set cross-sectional points 76 at the same set direction angle as the barycentric axis 75. The body outline line 78 is formed by setting a certain longitudinal section 77 and connecting the set section points 76 with lines on the longitudinal section 77. The longitudinal section 77 is, for example, a front side oblique direction with a set direction angle of −135 ° or −45 °, a side direction with a set direction angle of −180 ° or 0 °, and a set direction angle of 50 ° or 130 °. The body shape is set in four directions, that is, a diagonal direction on the back side and a front direction in which the set direction angle is −90 °, and the set cross-sectional points 76 are connected in the vertical direction for each longitudinal section 77. An outline line 78 is formed. The set section point 76 acquired in STEP: 03, the three-dimensional coordinates of the set section point 76, and the body outline line 78 acquired in STEP: 04 are associated with the set direction angle and the longitudinal section 77. And stored in the storage unit 28.

  When the longitudinal section 77 is set, the distance calculation unit 40 calculates a horizontal distance between the set section point 76 on the body outline line 78 and the barycentric axis 75, and totals the calculated distances. As shown in FIGS. 19 to 22, a graph is created with the vertical axis representing the distance r between the center of gravity axis 75 and the set cross-sectional point 76 and the horizontal axis representing the height. The envelope curve of the graph coincides with the body outline line 78.

  FIG. 19 is a graph 80 in which the horizontal distances between the center of gravity axis 75 and the set cross-sectional points 76 in the case where the set cross-sectional points 76 are located in the front side oblique direction of the body cross-sectional line group 69 are tabulated. FIG. 20 is a graph 81 that summarizes the horizontal distance between the center of gravity axis 75 and the set cross-sectional point 76 when the set cross-sectional point 76 is located in the oblique direction on the back side of the body cross-sectional line group 69. FIG. 21 is a graph 82 in which the horizontal distances between the center-of-gravity axis 75 and the set cross-sectional points 76 when the set cross-sectional points 76 are located in the lateral direction of the body cross-sectional line group 69 are summarized as FIG. It is a graph 83 in which the horizontal distances between the center of gravity axis 75 and the set cross-sectional points 76 when the set cross-sectional points 76 are located in the front direction of the body cross-sectional line group 69 are tabulated.

  In STEP: 03 and STEP: 04, the reference line 79 is rotated in a plurality of preset directions. However, the reference line 79 is rotated around the center of gravity X instead of the preset direction. The longitudinal section 77 where the body outline line 78 appears is appropriately selected, the body outline line 78 is acquired for each of the selected longitudinal sections 77, and the body outline line 78 and the direction angle are stored in association with each other. Also good. Further, the three-dimensional coordinates of the set cross-sectional points 76 forming the body outline line 78 at this time are also acquired.

  STEP: 05 When four types of graphs 80 to 83 are created for each direction in STEP: 04, the position of the neck part, the bust part, the waist part, and the hip part from the trunk part by the trunk part specifying unit 42 Is identified. The position of each part is specified using the graphs 80 to 83. For specifying the position by the graph, information indicating the characteristics of each part is used in the information such as the shape of the envelope curve, the change state of the distance r, the maximum value and the minimum value of the distance r, and the like. First, the body part specific part 42 is located when the set cross-section point 76 is located in the front side oblique direction of the body cross-section line group 69, that is, in the longitudinal section 77 extending in the front side oblique direction. The position of the neck part and the bust part is specified using the graph 80. Specifically, the body part specifying unit 42 specifies the section line 64 having the smallest distance r in the graph 80 as a neck dimension section line, and specifies the neck dimension section line as the position of the neck part.

  Further, as shown in the graph 80, when the set cross-section point 76 is located in the oblique direction on the front side of the body cross-section line group 69, the distance r is abruptly within a preset estimated bust height range. Is determined as a bust part, and the cross-sectional line 64 having the largest distance r in the bust part is determined as a bust part. The dimension cross section line is used, and the bust dimension cross section line is specified as the position of the bust part.

  After specifying the neck part and the bust part, the torso part specifying unit 42 uses the graph 81 in the case where the set cross-section point 76 is located in the diagonal direction on the back side of the torso cross-section line group 69 to determine the position of the hip part. Identify. That is, the torso part specifying unit 42 determines the lowermost mountain-shaped part as a hip part, and the cross-sectional line 64 having the largest distance r in the hip part is set as a hip dimension cross-sectional line, and the hip dimension is determined. The cross section line is specified as the position of the hip region.

  After specifying the hip part, the body part specifying part 42 specifies the position of the waist part using the graph 82 when the set cross-sectional point 76 is located in the side surface direction of the body cross-sectional line group 69. That is, when it is input that the body part of the measurement object 2 is thin, the body part specifying unit 42 is within an estimated waist height range set in advance between the neck part and the hip part. When the cross-sectional line 64 having the smallest distance r is defined as a waist dimension cross-sectional line and the body shape of the measurement object 2 is input to be thick, the distance r is the largest within the waist height range. The cross section line 64 is defined as a waist dimension cross section line, and the waist dimension cross section line is specified as the position of the waist portion.

  STEP: When the neck part, the bust part, the waist part, and the hip part are identified from the trunk section line group 69 in STEP: 05, the facial part identifying unit 43 next sets the set section point 76 to the trunk section line. The position of the facial part is specified using the graph 83 in the case where the group 69 is located in the front direction. The face part specifying unit 43 detects a nose part located above the specified neck part from the shape of the envelope curve of the graph 83, and the cross-sectional line 64 having the largest distance r among the detected nose parts. Is the nose dimension cross section line, and the nose dimension cross section line is specified as the position of the nose portion.

  Further, the face part specifying unit 43 determines a part located on the specified nose part as an eye part, and sets the cross-sectional line 64 having the smallest distance r in the eye part as an eye dimension cross-sectional line. The eye dimension sectional line is specified as the position of the eye part.

  Needless to say, the graphs 80 to 83 may be used not only for specifying each part but also for measuring the shape of the body part.

  (Step 07) When the position of the facial part is specified by the facial part specifying part 43, the automatic parting part 44 then specifies each part specified by the body part specifying part 42 and the facial part specifying part 43. Dimension data is automatically measured. That is, the dimension data of the neck part is acquired by measuring the circumference of the neck dimension section line, the dimension data of the bust part is acquired by measuring the circumference of the bust dimension section line, and the waist dimension section line Dimension data of the waist part is acquired by measuring the perimeter of the hip, dimensional data of the hip part is acquired by measuring the perimeter of the hip dimension cross section line, and the perimeter of the nose dimension cross section line is obtained. The dimension data of the nose part is acquired by measuring, and the dimension data of the eye part is acquired by measuring the perimeter of the eye part dimension cross section line. Each acquired dimension data is stored in the storage unit 28 in association with each part.

  STEP: 08 The blindfold processing unit 45 is a single color portion corresponding to the eye part specified by the face part specifying unit 43 in a three-dimensional image pasted with an image of white light irradiated to the TIN, For example, a model image filled in white is created, and the model image is displayed on the display unit 24, whereby the detailed part determination process by the second determination unit 27 is finished.

  After the detailed site determination process is completed, a table with a list of site names that have been subjected to automatic measurement processing such as a neck site, a bust site, and a waist site is displayed on the display unit 24. When the operator designates a desired part name via the operation unit 23, the dimension data of the designated part is displayed on the display unit 24.

  As described above, in the present embodiment, the noise data included in the TIN created from the images acquired by the imaging units 16 to 19 is converted into the first noise deletion processing unit 31 and the second noise deletion processing. Since the unit 33 and the third noise deletion processing unit 35 are configured to automatically delete, the three-dimensional shape of the measurement object 2 without noise can be measured easily and quickly without contact.

  Further, the cross-sectional line interpolation unit 36 and the three-dimensional shape interpolation unit 37 can interpolate the missing portion of the TIN generated by various noise deletion processes to create a TIN having no missing portion. By pasting the image irradiated with, a three-dimensional image without missing can be created.

  In the present embodiment, the first determination unit 26 can automatically recognize the human body as the measurement object 2 by classifying the human body into five parts of both arms, torso, and both legs, and the second determination unit 27. In particular, it is possible to automatically acquire dimensional data such as bust part, waist part, hip part, etc. in the body part with large individual differences, so that measurement data can be easily used in a wide range according to the application such as clothing production. Can do.

  Further, a plurality of different longitudinal sections 77 having directional angles set in advance, including the barycentric axis 75 and the set section point 76, and each set section point 76 positioned on each longitudinal section 77; Since the graphs 80 to 83 are created by counting the horizontal distance from the center of gravity axis 75, the graph to be used can be selected according to the shape of each part, and erroneous detection of each part is prevented. Can do.

  Moreover, since the dimension data of the designated part is displayed on the display unit 24 simply by designating a desired part from the table displayed on the display unit 24, the operator can easily The dimension data of a desired part can be easily acquired by operation.

  In addition, by digitizing one's body shape, it can be used for health management and the like by comparison with past numerical values.

  Further, the face part specifying unit 43 automatically specifies the eye part and fills the part corresponding to the eye part of the three-dimensional image with a single color, so that the individual is specified by the three-dimensional image. Can be difficult, and privacy can be protected.

  In this embodiment, the dimension data of the designated part is displayed on the display unit 24 by designating the desired part from the table in which the list of the part names where the measurement was performed is described. However, the dimension data of the part that has been subjected to the automatic measuring process together with the model image may be automatically displayed on the display unit 24.

  Further, rather than setting the direction angle in which the longitudinal section 77 is formed in advance, a plurality of the longitudinal sections 77 are created at an arbitrary angle, and the set section point 76 existing on each longitudinal section 77; By creating a graph summing up the horizontal distance between the centroid axes 75 for each vertical section 77, when the operator designates a desired part from the model image, the graph for each vertical section 77 is searched. Then, the graph having the body outline line 78 that is the most characteristic for the designated part is selected, the position of the designated part is specified based on the selected body outline line 78, and the dimension data is acquired. You may do it.

  In addition, instead of designating a desired part from a table displaying a list of part names, the model image is displayed on the display unit 24, and a desired part is designated in the image via the operation unit 23. Thus, the dimension data of the designated part may be displayed on the display unit 24.

  Moreover, although the said measuring object 2 was demonstrated as a human body in the said Example, you may implement to three-dimensional measurement of crafts, such as a bag.

  In this embodiment, a horizontal section is set in the vertical direction in the three-dimensional shape data at predetermined intervals, and the cross-sectional line 64 is created based on the horizontal section. A vertical cross section may be set at a predetermined interval, a vertical cross section line may be created based on the vertical cross section, and the vertical cross section lines may be stacked to create a vertical cross section line group. In the case of the vertical cross-section line group, as in the case of the body cross-section line group 69, a center-of-gravity axis is created from the average center-of-gravity coordinates obtained based on the center of gravity of each vertical cross-section line. By rotating the reference line in a plurality of directions and acquiring the intersection of the reference line and the vertical cross-section line as the set cross-section points, respectively, the cross-sections that are planes including the center of gravity axis and the set cross-section points in different directions The vertical distance between the set cross-sectional point and the center of gravity axis is calculated and aggregated to create a plurality of graphs, and a desired part is specified based on the plurality of graphs.

DESCRIPTION OF SYMBOLS 1 Three-dimensional measuring system 2 Measurement object 16-19 Imaging unit 22 Arithmetic controller 25 Arithmetic control part 26 1st determination part 27 2nd determination part 29 Matching process part 31 1st noise deletion process part 32 Parts recognition process part 33 1st 2 Noise Deletion Processing Unit 34 Cross Section Line Creation Unit 35 Third Noise Deletion Processing Unit 36 Cross Section Line Interpolation Unit 37 3D Shape Interpolation Unit 38 Center of Gravity Axis Creation Unit 39 Cross Section Point Coordinate Acquisition Unit 40 Distance Calculation Unit 41 Site Specification Unit 42 Torso Site Identification part 43 Face part identification part 44 Automatic measurement processing part 45 Blindfold processing part

Claims (9)

  A plurality of imaging units that have two cameras for taking a stereo image of a measurement object, and feature points are extracted from the two images obtained by the imaging units, and the feature points that are common to the images are three-dimensional. A three-dimensional measurement system comprising: an arithmetic control device that measures and measures a three-dimensional shape of the measurement object for each imaging unit and creates a three-dimensional image of the measurement object, the plurality of imaging units Is arranged so as to image the measurement object from a plurality of directions, and the arithmetic and control unit creates three-dimensional shape data of the measurement object based on the TIN created based on the feature points, and the three-dimensional shape Horizontal cross sections are set at predetermined intervals in the vertical direction in the data, cross section points that are intersections of the horizontal cross sections and the triangles forming the TIN are created to create cross section lines, and the cross section lines are stacked at predetermined intervals. A first determination unit that creates a group of cross-sectional lines, and obtains a centroid position of each cross-sectional line and obtains a centroid axis based on an average of the centroid positions. A point is set as a set cross-sectional point, a vertical cross-section including the set cross-sectional point and the center of gravity axis is set, and a body outline line is created by connecting the set cross-section points in the vertical cross section, and the body outline line is used as a basis. And a second determination unit that determines a position of a predetermined part.   The second determination unit creates a plurality of the body contour lines by setting a plurality of the longitudinal cross sections in different directions and a section point coordinate acquisition unit that acquires the three-dimensional coordinates of the set section points, and the body contour A distance calculation unit that calculates and aggregates the horizontal distance between the set cross-sectional point on the line and the centroid axis for each body outline line and creates a plurality of graphs, and a detailed part of the measurement object based on the created graphs The three-dimensional measurement system according to claim 1, further comprising: a part specifying unit that determines the position.   The three-dimensional measurement system according to claim 2, wherein the part specifying unit specifies a dimension cross-sectional line of a predetermined part based on one graph selected from the plurality of graphs.   The three-dimensional measurement system according to claim 3, wherein the part specifying unit specifies a dimension cross-sectional line of another part from the cross-sectional line group based on the position of the specified part.   The three-dimensional measurement system according to claim 4, wherein the second determination unit further includes an automatic measurement processing unit, and the automatic measurement processing unit measures the circumference of the specified dimension cross-section line as dimension data.   A display unit that displays a table showing a list of site names created by the arithmetic and control unit; and an operation unit that can specify a desired site from the table displayed on the display unit. The three-dimensional measurement system according to claim 5, wherein the dimension data of the part is displayed on the display unit by designating a desired part.   A display unit that displays a three-dimensional image created by the arithmetic and control unit; and an operation unit that can specify a desired part of the three-dimensional image. The desired part of the three-dimensional image by the operation unit. The three-dimensional measurement system according to claim 5 or 6, wherein the dimension data of the part is displayed on the display unit.   A step of taking a stereo image of a measurement object with a plurality of imaging units having two cameras, extracting feature points in a plurality of images obtained by the plurality of imaging units, and measuring the measurement object based on the feature points A horizontal section at a predetermined interval in the vertical direction on the TIN, obtaining a cross-section point that is an intersection of the horizontal cross-section and the triangle constituting the TIN, and A step of creating a cross-sectional line by connecting in a horizontal cross-section, a step of creating a cross-sectional line group by laminating the cross-sectional lines at a predetermined interval, a step of obtaining a barycentric position of each cross-sectional line and creating a barycentric axis; Obtaining a point of each cross-sectional line in a predetermined angle direction around the center of gravity position as a set cross-section point, setting a vertical cross section including the set cross-section point and the barycentric axis, and setting the set cross-section point to the vertical cross section Outside the body by connecting in Three-dimensional measuring method characterized by comprising the step of creating a line, and identifying a position of a predetermined region based on said body type contour line.   In a three-dimensional measurement system having a plurality of imaging units having two cameras and an arithmetic control device that controls the imaging units, the arithmetic control device causes the plurality of imaging units to photograph a measurement object in stereo, and Extracting feature points in a plurality of images obtained by a plurality of imaging units, creating a TIN of the measurement object based on the feature points, setting a horizontal section at a predetermined interval in the vertical direction in the TIN, A cross-section point that is an intersection of the horizontal cross-section and the triangle that constitutes the TIN is obtained, and the cross-section points are connected in the horizontal cross-section to create a cross-section line, and the cross-section lines are stacked at a predetermined interval. A line group is created, the center of gravity of each cross-sectional line is obtained, the center of gravity is created, the point of each cross-sectional line in a predetermined angle direction around the barycentric position is obtained as a set cross-sectional point, Heavy A vertical cross section including an axis is set, a body outline line is created by connecting the set cross section points in the longitudinal section, and a position of a predetermined part is specified based on the body outline line 3D measurement program.