JP2000241135A - Method of detecting contour of object and method of identifying object - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a contour detecting method whereby the contour of an object can be accurately detected, irrespective of the surface shape of the object and even if a slit light is used. SOLUTION: In a method of detecting the contour of an object 2 mounted on a conveyer 3, a plurality of slit lights 5 are projected from above the conveyer 3 so that the respective projection lines array like one line on the conveyer 3 as seen from right and left, a camera 8 photographs a specified range including this one line, say, reference projection line, the image of an object 2 taken by the camera 8 is recorded as binary data in an image memory 2 composed of pixels sectioned by a plurality of longitudinal lines parallel to the reference projection line and a plurality of transverse lines perpendicular to the reference line, the difference between data of the transverse pixel columns in the image memory and data of the transverse reference pixel columns obtained by photographing the reference projection line in the image memory is obtained, and the position on a pixel column where this difference exceeds a preset value is determined to be the contour of the object.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an object outline detecting method for detecting an outline of an object placed on a plane, and an object identifying method for identifying the object.

[0002]

2. Description of the Related Art Heretofore, in the case of automatically detecting the shape of an object, a camera device has been used and a predetermined image processing has been performed on the photographed image to obtain, for example, its outline. Specifically, light is irradiated from above the object and the whole is photographed by a camera device or the like, or an operation of irradiating slit light and detecting the outer shape of a part of the whole is performed, and the contour of the object is Had been detected.

[0003]

As described above, in the method of photographing the entire object by the camera device, the contrast of the object depends on the surface properties of the object and the lighting conditions. When there are characters or patterns on the surface, or when the shadow of the object is projected on the background, there is a problem that the contour of the object cannot be accurately extracted.

In addition, when slit light is used, a shadow is generated depending on the shape of the object at any position of the light source, and a boundary point of the object is erroneously detected. There was a problem that it could not be detected. Therefore, the present invention provides a surface property of an object,
It is an object of the present invention to provide an object contour detection method and an object identification method that can detect the contour of an object regardless of the surface shape and can accurately detect the contour even when slit light is used.

[0005]

According to a first aspect of the present invention, there is provided a method for detecting an outline of an object placed on a plane, the method comprising the steps of: A plurality of slit light beams are projected from above onto the plane such that the respective projection lines overlap with one another, and a predetermined range including the reference projection line overlapping with the one is captured by the camera device, In addition, when the camera device photographs an object on a plane, a position where a plurality of surface projection lines generated on the surface of the object are displaced from the reference projection line is regarded as an outline of the object.

According to the above-described method for detecting the contour of an object, a plurality of slit light beams are projected on the object, the surface of the object is photographed, and the contour is detected. Blind spots hardly occur compared to the case where
The reliability of detection for detecting the contours of a plurality of objects increases.

A second object contour detecting method according to the present invention is a method for detecting the contour of an object placed on a plane, wherein a plurality of slit light beams are projected from above the plane. The lines are projected onto the plane so as to overlap with one another, and a predetermined range including the reference projection line overlapping with the single line is photographed by a camera device. The image data is recorded as binary data in an image recording unit composed of a plurality of pixels partitioned by a plurality of vertical lines parallel to the line and a plurality of horizontal lines perpendicular to the reference projection line. The difference between the data of each pixel row and the data of the reference pixel row in the horizontal direction obtained by photographing the reference projection line is obtained, and the position in the pixel row where the difference exceeds the set value is used as the contour of the object. is there.

According to the above-described contour detection method, upon detecting a contour, an image obtained by photographing a surface projection line generated on the surface of an object and a reference projection line generated on a plane is recorded as binary data. Since the contour is detected based on the difference, only a very simple calculation is required.
Therefore, it is very suitable for processing in real time, and can accurately detect the contour of the object regardless of the surface properties and surface shape of the object.

Further, the object identification method according to the present invention is a method for identifying an object placed on a plane, wherein a plurality of slit rays from above the plane are overlapped by one. In this manner, a predetermined range including the reference projection line overlapped with one of the projection lines is projected by the camera device, and an image captured by the camera device is transferred to a plurality of parallel images parallel to the reference projection line. Is recorded as binarized data in an image recording unit composed of a plurality of pixels divided by a plurality of horizontal lines perpendicular to the vertical line and the reference projection line, and data of each pixel row in the horizontal direction in the image recording unit is Obtain the difference from the data of the horizontal reference pixel row obtained by capturing the reference projection line, consider the position in the pixel row where the difference exceeds the set value as the contour of the object, and Project objects based on contours Together determine the area, by comparing the with the specified area size value this projected area is a method for identifying whether a predetermined object.

According to the object identification method, the projection area of the contour of the object detected by the same method as the object contour detection method is determined, and the projection area value is calculated in advance with the projection of the predetermined object to be identified. Since the comparison is made with the set area value which is the area, the predetermined object can be easily and reliably identified.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an object contour detecting method and an object identifying method according to an embodiment of the present invention will be described with reference to FIGS.
It will be described based on. In this embodiment, for example, in order to detect waste that cannot be reused from waste (hereinafter, referred to as unsuitable), the waste, which is an object, is placed on a transport body and moved. A case will be described in which an image is captured by a camera device, the outline is detected, and whether or not the object is inappropriate is identified based on the outline.

First, an object detection / identification device for detecting an outline of an object and identifying an inappropriate object based on the outline will be described. As shown in FIGS. 1 and 2, the object detection / identification device 1 includes a carrier that moves an object 2 along a predetermined transport path K, for example, a belt conveyor (hereinafter, referred to as a belt conveyor).
A support base 4 erected above a conveyor 3; and a plurality of, for example, six (of course, not limited to six) And a projector 7 for projecting the slit light beam 5 onto the conveyor 3 from the front and rear in the transport direction such that the respective projection lines 6 overlap with one another, and provided above the support base 4. A camera device 8 using a CCD for photographing a predetermined range (for example, within a fixed left and right distance of the reference projection line and also referred to as a detection area) including the single reference projection line 6S; An image processing device (computer device) 9 for inputting an image captured by the camera device 8 to detect the contour of the object 2 and to determine whether the object 2 is an unsuitable object other than recyclable garbage. And composed from To have. The camera device 8
Is usually provided directly above the reference projection line 6S, and the slit light beam 5 is projected from both sides thereof (the front and rear sides in the transport direction, and the left and right sides when the reference projection line is the center). Generally speaking, the relationship between the camera device 8 and the reference projection line 6S only needs to be such that at least a plurality of slit rays 5 are projected from both sides of a plane connecting the camera device 8 and the reference projection line 6S.

As shown in FIG. 3, the light projector 7 includes, for example, a straight tube fluorescent lamp 11 as a light source, and the straight tube fluorescent lamp 11.
A polarizing lens 12 disposed below the polarizing filter 12 for polarizing the light into parallel rays, and a convex lens 1 disposed below the polarizing filter 12 and emitting six slit rays 5
3 and the like. The arithmetic processing unit 9 includes:
Although not shown, a contour detection unit for detecting the contour of the object and an object identification unit for identifying whether or not the object is an unsuitable object other than recyclable waste based on the contour detected by the contour detection unit are provided. Have been.

The outline detecting section compares the data recorded in the image memory (image recording section) with data obtained by photographing only the reference projection line, that is, the reference projection. A comparison detection unit that compares the line with a surface projection line projected on the surface of the object, extracts a boundary point of the object, and detects an outline. FIG.
As shown in the figure, in the image memory 21, the image of the photographed object is represented by m columns × horizontal direction (direction parallel to the object conveyance direction) in the vertical direction (direction orthogonal to the object conveyance direction).
Is obtained by being divided into a large number of pixels [pixels (hereinafter, referred to as "pix") 22 composed of n rows, and each of the pixels 22 has a binary Data, that is, data of “1” and “0” is recorded (stored).

For example, as shown in FIG. 5, a plurality of surface projection lines 6P, which are projections of the slit ray 5 projected on the object 2 on the conveyor 3, appear along the surface at each position. This image captured by the camera device 8 is
In the state as shown in FIG. That is, the plurality of slit rays 5 in the image memory 21 are converted into one reference projection line 6S on the conveyor 3 having no object, and
At the position deviated from the projection line 6P, the memory corresponding to each pixel stores the projection line 6P.
"1" and "0" data are recorded according to the presence or absence of.

The column width of the image memory 21 in the horizontal direction is, for example, a predetermined multiple of the width of the reference projection line, specifically, at least about five times. In addition, the comparison and detection unit obtains a difference between an image obtained by capturing the plurality of surface projection lines 6P appearing on the surface of the object 2 and an image obtained by capturing only the reference projection line 6S, that is, a difference between binary data. An arithmetic circuit is provided.

In this arithmetic circuit, a data row (1 × m pixels) of a horizontal pixel row in the image memory 21 is regarded as an m-dimensional vector, and only the reference projection line 6S is obtained as shown in the following equation (1). Is a data row of one pixel row (reference pixel row) that captures the image, and a shooting vector f _i (i = 1 to _m ) that is a data row of each pixel row that captures a plurality of surface projection lines 6P. ) (Which is represented by the total absolute value of the difference between the components of the vector and is also referred to as pattern matching) S is determined, and the Euclidean distance S is compared with a predetermined threshold (set value). . For example, a numerical value of “1” or more is set as the threshold.

[0018]

(Equation 1) More specifically, as shown in FIG.
P of the i-th row of shooting vector value f _i is (00001000
If it is 0), the reference vector f _x also (00001
0000), the Euclidean distance S _i becomes zero. However, when the vector value f _k of the k-th row of the surface projection line 6P is (011000101), subtracting the reference vector f _x (000000100000) gives Since the Euclidean distance S _k has different values at five places,
It becomes “5”.

Then, as shown in FIG.
Each Euclidean distance S obtained_iIs a preset threshold
Vertical coordinates of the first pixel row that is above ("1" or more)
Value y _h, And the vertical coordinate value y of the last pixel row_lGet
These values are used as boundary points 23 indicating the outline of the object 2.
Extracted. Therefore, in the above-described contour detection unit,
The extraction operation of the boundary point is performed every fixed sampling period T
At a predetermined interval (conveyor speed V × T)
And by repeating over the entire length of the object,
As shown in FIG. 7, when each boundary point 23 is obtained
By connecting these boundary points 23 together,
Thus, an outline 24 of the object 2 is obtained.

As described above, a plurality of slit light beams 5 are projected on the surface of the object 2 so as to overlap one projection line on the conveyor 3 on which the object 2 is placed and moving. An image obtained by photographing the surface of the object 2 by the camera device 8 is recorded as binary data in an image memory 21 composed of a plurality of pixels divided vertically and horizontally, and is stored in each pixel row in one horizontal row of the image memory 21. Data and
By calculating the Euclidean distance with the data of the horizontal row of pixel rows taken only of the reference projection line, and comparing whether this Euclidean distance is equal to or greater than a predetermined threshold,
The boundary point 23 of the object 2 is extracted, that is, the contour 2 of the object 2 is extracted.
4, the surface properties of the object 2,
The contour 24 can be reliably detected regardless of the surface shape. In addition, since a plurality of slit light beams are projected from the left and right, a blind spot hardly occurs compared to, for example, the case of using a single slit light beam. For detecting the contour of a certain object, the detection reliability is improved.

In detecting the contour, an image obtained by photographing a surface projection line generated on the surface of the object and a reference projection line generated on a plane is recorded as binary data.
Since the contour is detected based on the difference between the two, very simple calculations are required, and therefore, it is very suitable for processing in real time. In addition, the lighting conditions for the object, the surface properties of the object, and the surface Regardless of the shape,
If the contrast between the object and the background is unclear,
Also, even when there are characters or patterns on the surface of the object, the contour of the object can be detected accurately.

Next, the object discriminating section includes an area calculating section for calculating a projection area (for example, obtained from the number of pixels) S of the object 2 based on the contour 24 obtained by the contour detecting section; Width L corresponding to the maximum fillet diameter at
(See FIG. 7), and a determination for determining whether or not the object 2 is inappropriate based on the projection area S and the maximum width L obtained by the two calculation units. Unit is provided.

As shown in FIG. 7, for example, the area calculating section calculates the coordinate values of the upper and lower boundary points 23a and 23b obtained in the previous sampling (in the direction orthogonal to the transport direction) and the current sampling point. Upper and lower boundary points 2 obtained by
The projection area S of the object 2 is obtained by calculating the trapezoidal partial area by the area formula using the coordinate values of 3c and 23d, and adding this partial area over the entire length of the object.

In the width calculating section, each boundary point 2
The maximum width L in the outline 24 is obtained from the coordinate value of No. 3. Therefore, the maximum width L corresponding to the maximum fillet diameter of the object 2 can be easily obtained. Further, the determination unit determines whether or not the object is inappropriate based on the projection area S and the maximum width L obtained by each of the calculation units.

Specifically, as shown in the graph of FIG.
In advance, the projected area S and the maximum width L of the recyclable garbage (cans, glass bottles, PET bottles, etc.) and unsuitable objects contained in the conveyed waste are determined from experiments,
Then, by plotting these data on a graph, a value that can discriminate the recyclable waste from the unsuitable material, for example, a projected area value is extracted.

For example, as can be seen from the graph of FIG. 8, the area S of an unsuitable object indicated by a black mark is 4 × 10 ⁴ pi
With a value of x ^{2 or more} (a square pixel, for example, one square pixel is about 1 mm ² , and therefore is equivalent to 4 × 10 ⁴ mm ² ), everything is inappropriate and the maximum width L, which is the maximum fillet diameter, is also , 300 pix (in this case, 300 m
m), if it is less than one bottle, beer bottle, PE
Excluding T bottles, the value is larger than that of most recyclable waste.

Therefore, the projected area S of the object 2 to be conveyed is 4 × 10 ⁴ pix ² or more and the maximum width L is 30
If it is 0 pix or less, it is determined to be inappropriate, that is, identified. As described above, at least the projected area obtained based on the detected contour of the object is compared with a set area value which can be determined in advance as an unsuitable object by an experiment or the like, so that it can be easily performed. In addition, it is possible to accurately determine whether or not the object is inappropriate.

As the light source of the projector, monochromatic light of a specific wavelength is used, and an optical filter capable of receiving only light of this wavelength is mounted on the camera device. By the way, in the above embodiment, a polarizing filter,
Although a convex lens or the like is used, for example, as shown in FIG. 9, a plurality of slit light sources 31 are installed with their emission directions changed so that the tips of the emitted slit light beams 32 are gathered at one place. Such a configuration may be used.

In the above embodiment, waste is described as an object. However, the present invention can be applied to, for example, detecting or identifying a person or a vehicle on a road in order to check a traffic amount.

[0030]

As described above, according to the object contour detecting method of the present invention, a plurality of slit light beams are projected on the object to photograph the surface of the object and detect the contour. As compared to the case where a single slit beam is used, there is almost no blind spot, and therefore, detection of objects having a complicated shape or contours of a plurality of objects has a high detection reliability. Get higher.

In detecting the contour, an image obtained by photographing the surface projection line generated on the surface of the object and the reference projection line generated on the plane is recorded as binary data.
Since the contour is detected based on the difference between the two, very simple calculations are required, and therefore, it is very suitable for processing in real time. In addition, the lighting conditions for the object, the surface properties of the object, and the surface Regardless of the shape,
If the contrast between the object and the background is unclear,
Also, even when there are characters or patterns on the surface of the object, the contour of the object can be detected accurately.

Further, the projection area of the contour of the object detected by the same detection method as described above is obtained, and this projection area value and a preset area value which is a predetermined projection area of a predetermined object to be identified are determined. Are compared, a predetermined object can be easily and reliably identified.

[Brief description of the drawings]

FIG. 1 is a side view showing a schematic configuration of an apparatus for detecting and identifying an outline of an object according to an embodiment of the present invention.

FIG. 2 is a perspective view showing a schematic configuration of a light projector in the same device.

FIG. 3 is a side view showing a main configuration of the light projector.

FIG. 4 is a plan view showing a configuration of an image memory in the device.

FIG. 5 is a plan view showing a projection line when the image is projected on an object in the apparatus.

FIG. 6 is a plan view showing an image memory of a captured image in the apparatus.

FIG. 7 is a plan view showing an outline of an object photographed by the device.

FIG. 8 is a graph illustrating a relationship between a maximum width of an object and a projected area for describing an object identification method according to an embodiment of the present invention.

FIG. 9 is a side view showing a configuration of a main part of a modification of the light projector in the same device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Object detection identification apparatus 2 Object 3 Belt conveyor 5 Slit ray 6 Projection line 6S Reference projection line 6P Surface projection line 7 Projector 8 Camera device 21 Image memory 22 Pixel 23 Boundary point 24 Contour

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 2F065 AA12 AA22 AA51 AA58 AA61 CC00 DD06 DD11 FF01 FF04 GG03 GG22 HH05 HH12 HH14 JJ03 JJ09 JJ26 LL22 LL32 MM03 PP15 QQ01 QQ04 QQ24 QQ25 QQ27 QQ38

Claims (3)

[Claims] 1. A method for detecting an outline of an object placed on a plane, comprising: projecting a plurality of slit light beams from above the plane onto the plane such that respective projection lines overlap with one another. A plurality of surfaces formed on the surface of an object when projected and photographed in a predetermined range including the reference projection line overlapped by the camera device, and when the camera device photographs an object on a plane. An object contour detection method, wherein a position where a projection line is displaced from a reference projection line is regarded as an object contour. 2. A method for detecting an outline of an object placed on a plane, comprising: projecting a plurality of slit light beams from above the plane onto the plane such that respective projection lines overlap with one another. The camera device projects a predetermined range including the reference projection line overlapped with one of the projection lines. The image taken by the camera device is converted into a plurality of vertical lines parallel to the reference projection line and the reference projection line. It is recorded as binarized data in an image recording unit composed of a plurality of pixels divided by a plurality of horizontal lines perpendicular to the line, and the data of each pixel row in the horizontal direction and the reference projection line in this image recording unit are photographed. A method for determining a difference from data of a reference pixel row in the horizontal direction obtained as described above, and using a position in the pixel row in which the difference exceeds a set value as an object contour. 3. A method for identifying an object placed on a plane, comprising: forming a plurality of slit light beams from above the plane;
Each projection line is projected onto the plane so as to overlap with one line, and a predetermined range including the reference projection line overlapping with the one line is photographed by a camera device, and an image photographed by the camera device is taken. The image is recorded as binary data in an image recording unit composed of a plurality of pixels partitioned by a plurality of vertical lines parallel to the reference projection line and a plurality of horizontal lines perpendicular to the reference projection line. The difference between the data of each pixel row in the horizontal direction and the data of the reference pixel row in the horizontal direction obtained by photographing the reference projection line is determined, and the position in the pixel row where the difference exceeds the set value is defined as the contour of the object. Deemed, and determining the projected area of the object based on the obtained contour of the object, and comparing the projected area with a set area value to identify whether or not the object is a predetermined object. Object identification method