JPH03139776A - Object identifying method - Google Patents

Abstract

PURPOSE:To shorten the signal processing time of high identification information while maintaining the information by identifying and judging the object by using a binary image of a brightness signal in a 1st identifying process and identifying and judging the object by using a binary image of chromatic components in a 2nd identifying process when the object can not be identified in the 1st identifying process. CONSTITUTION:An external equipment I/F 36 is connected to a CPU 18 through a data balance line 28 and signals indicating the decision result of kind identification, the result of a position measurement, etc., are outputted from its external equipment I/F 36 to a classifying device,n automatic working machine, etc., through a parallel I/F 38 and a serial I/F 40. When a visual device identifies the kind of a component 10, the identification processing is performed by using the binary image of the brightness signal SY. Then the binary image of chromatic components SR, SG, and SB is used when necessary. Consequently, while high identification accuracy is maintained, its signal processing time is shortened as compared with a case wherein signal processing is performed by using the chromatic components SR, SG, and SB from the beginning.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to an improvement in an object identification method for identifying the type of object based on a color image signal.

2. Description of the Related Art In conventional automatic processing machines, the type of object is identified using an imaging device such as a TV camera, and a process corresponding to the type is automatically performed. For example, JP-A-6
The device described in Japanese Patent No. 3-142791 is an example of such a device, which identifies the type of object based on a binary image based only on brightness. Such a processing device has the advantage that the signal processing time is fast and the device can be constructed simply and inexpensively, but on the other hand, it has the disadvantage that it is not possible to distinguish between objects that have the same or similar shapes.

In response to this, a device has been proposed that identifies the type of object based on a color image signal output from a color imaging device such as a color camera. According to this method, not only the shape but also the color is determined, so that even if the shapes are the same, the type of object can be identified with high accuracy based on the difference in color.

Problems to be Solved by the Invention However, conventional image processing apparatuses using such a color imaging device have been used for R (red) and G (green).

Using the color data of the three primary colors of B (blue), divide it into, for example, 8 bits: 256 gradations and digitize it, store each data in a frame memory, and read out the data for discrimination processing. As a result, the amount of data to be processed is extremely large, requiring a large-capacity storage device, making the device expensive and prolonging the processing time.

The present invention has been made against the background of the above circumstances, and its purpose is to enable the type of object to be identified at high speed and with high precision using color image signals.

Means for Solving the Problems In order to achieve this objective, identification processing is first performed based on the luminance signal of the color image signals, and when identification cannot be performed using only the luminance signal, identification processing is performed based on the color signal. The present invention can identify which of a plurality of predetermined types of objects an object imaged by a color imaging device is based on a color image signal output from the color imaging device. An object identification method comprising: (a) a first identification step of identifying the type of the object using a binary image obtained based on a luminance signal related to light intensity among the color image signals; ) When the type of the object cannot be identified in the first identification step, using a binary image regarding the color of the object obtained based on color signals regarding a plurality of hues among the color image signals. A second identification step for identifying the type of the object.

In addition, in the second identification step, it is not necessary to make identification judgments regarding all types of objects; for example, it is not necessary to make identification judgments regarding all types of objects.
Or, for only a few types of objects, there are two
The identification judgment may be made based on the value image.

Operation and Effects of the Invention In such an object identification method, first, in a first identification step, a binary image based on a luminance signal is used to identify an object. In this case, the object is identified in the second identification step using a binary image based on color signals, so for example, for an object with a characteristic shape, identification is completed only in the first identification step; When there are multiple types of objects with the same or similar shapes, a second identification step is performed to perform identification that takes into account the color of the object, and the signal is processed while maintaining high identification accuracy. Processing time is reduced.

In particular, in both the first and second identification steps, identification judgments are made based on binary images. The amount of data can be extremely small, the signal processing time can be significantly shortened, and a large-capacity frame memory or the like can be eliminated, making it possible to construct the apparatus at low cost.

In addition, in this way, both the first and second identification steps are performed twice.
Since the identification process is performed using value images, it is possible to share the hardware and software for performing identification processing on binary images, and the apparatus can be configured more simply and at low cost.

EXAMPLE Hereinafter, an example of the present invention will be described in detail based on the drawings.

FIG. 1 is a block diagram illustrating the configuration of a visual device for identifying object types according to the method of the present invention. In FIG. 1, reference numeral 10 is a part as an object supplied to a sorting device, an automatic processing machine, etc., and is one of a plurality of types of objects having different colors and shapes. A color camera 12 as a color imaging device is installed above the component 10, and receives the light reflected by the component 10 and outputs a luminance signal SY representing the light intensity and the color of the light R (red). , G (green), B
Color signals SR and SG are represented by three primary colors (blue).

A color image signal SP synthesized with SB is output.

The color image signal SP is supplied to a separation circuit 14, where it is separated into the luminance signal sy and color signals SR, 5O3B. All these signals are analog signals, and the luminance signal S
Y is supplied to a binarization comparator 16. The binarization comparator 16 is supplied with a level signal SL output from the CPU 18, which does not represent the binarization level, via the D/A converter 2o. The signal SY is compared with the analog level signal SL, and the luminance signal SY is converted into the level signal S.
A binary signal SY” expressed by two values depending on whether it is above or below L.
Convert to The level signal SL is set in consideration of illuminance and the like so that the luminance signal SY can be divided into those reflected by the component 10 and those not reflected.

On the other hand, the color signals SR, SG, and SB are analog signals related to a plurality of hues, and the extraction circuits 22R and 22 respectively
G, 22B. These extraction circuits 22R, 2
2G, 22[1 also have data signals SDR, SDG, S corresponding to color data set in advance from the CPU 18.
Dl] are supplied via D/A converters 24R, 24c, and 24n, respectively, and the extraction circuit 2
2R122G, 22++ are the color signals SR, SG, SB supplied to each pixel and the analog data signal SDR.
, S Dc , and S DR, respectively, and output only signal components within the range of the data signals SDR, SD, , SD. Data signals SDR, SDc, SD
e each have a certain width, and the extraction circuits 22R,
22G and 22B are configured to include, for example, a comparator that passes a signal component whose width is lower than the upper limit thereof, and a comparator that passes a signal component whose width is higher than the lower limit.

The color signals SR, SG, and SB that have passed through the extraction circuits 22R, 22, and 22Il are supplied to a color image detector 26. The color image detector 26 outputs a binary signal SC representing each pixel with two values depending on whether all the color signals SR, SG and SB are supplied. Such a color image detector 26
is configured to include, for example, an AND circuit that outputs a signal only when all of the color signals SR, SG, and SB are supplied.

The CPU 1B is connected to a ROM 30 . RAM32 and video RAM
34 constitutes a microcomputer, and RA
Signal processing is performed according to a program stored in advance in the ROM 30 while utilizing the temporary storage function of the M32 and the video RAM 34, and the level signal SL and data signal S are
In addition to outputting D, SDc, and SDR, 2(+!
Signals sy' and SC are read in via data bus line 28. The binarization level represented by the level signal SL is:
The information is stored in advance in the ROM 30 by an external device such as a personal computer connected via a setting device or connection terminal (not shown). In addition, data signals SDR, SD, , SD
The color data represented by , is stored in advance in the ROM 30 by a setting device or an external device, along with characteristic data such as the area of the object, the number of holes, the circumference length, etc., for each type of object. This color data represents the color of the object using the color signals SR and 5G.
Data of each signal value when expressed by SB,
More specifically, it represents the shading, or mixing ratio, of each color when mixing the three primary colors of red, green, and blue to create the color of the object, and it varies depending on changes in lighting, individual differences in the object, etc. A certain width is set in consideration of slight differences in color, etc.

In addition, the above two plant numbers SY" and SC are each part 1.
It represents the shape of 0 and is stored in video RAM3 as a 2-line image
4 and read out by the CPU 18 to identify its type, measure its position, etc. Position measurement is performed, for example, based on a binary image based on the above-mentioned 2 plant type SY", and type identification is performed, for example, by comparing the feature data stored in the ROM 30 with the above-mentioned binary image. This is carried out according to the flowchart 1 shown in the figure.

That is, first, in step S1, the luminance signal SY
A binary image based on the binary signal SY° obtained based on The type is identified. Specifically, the degree of coincidence between each feature data of a plurality of types of objects stored in advance and the binary image is calculated using SRI.
Each is calculated as a statistical distance according to an algorithm. The value of this statistical distance becomes closer to zero as the degree of coincidence between the feature data of the object and the binary image increases.

Then, in step S3, it is determined whether or not the type of the component 10 has been correctly identified, that is, whether or not the type of the component 10 is identified as one based on the identification result.
If it is specified, the series of signal processing related to type identification ends. The determination method in step S3 is as follows:
Extract three types of objects with small statistical distance values,
First, a first check is performed depending on whether the first statistical distance with the smallest value is smaller than a predetermined constant value, and then the first statistical distance and the second statistical distance are compared. A final determination is made by performing a second check depending on whether the deviation from the third-rank statistical distance is larger than a predetermined constant deviation value. Specifically, if the first statistical distance is smaller than the above-described - constant value, and the deviation from the second and third statistical distances is larger than the above-described certain deviation value, the component 10 It is determined that the object is the object for which the first statistical distance has been calculated, and the series of signal processing ends.

However, if the first statistical distance is larger than the above-mentioned constant value, or if the deviation from the second and third place statistical distances is smaller than the above-mentioned fixed deviation value even if it is smaller than the certain value, Step 34 and subsequent steps are executed without specifying the type of component 10.

In step S4, data signals SD, , SDG, and SD of color data regarding a plurality of types of objects are output, and two system codes SC regarding each color are read and temporarily stored in the video RAM 30. This data signal SD,,SD.

The output of and SD and the reading of the two plant numbers SC do not necessarily need to be performed for all objects, and in this example, the three types of objects with small statistical distances extracted in step S3, 1 2 or only for the first and second ranked objects within the predetermined deviation value range.

Here, when the color of the component 10 and the color represented by the data signals SD, SDr, and SD n match, the signal values of the color signals SR, SG, and SB are the same as those of the data signals SDR, SDG, and Within the range of SDR, part 10
A two-set signal SC corresponding to the shape of is output from the color image detector 26, but the color of the part 10 and the data signals SDR, SD
G and S.D.

If the color represented by is different from the color represented by SR, three color signals SR are used.

At least one of the signal values SG and SB falls outside the range of the data signals SD, l, SD, and SD, and the color image detector 26 outputs a two-signal signal SC consisting of only one of the two values. Therefore, a binary image corresponding to the shape of the component 10 cannot be obtained.

Then, in step S5, the binary image based on the two system codes SC stored in the video RAM 30 is read out, and in step S6, the data signals SDa, SDc and SDR output when obtaining the binary image are read out.
The feature data of the object and its binary image are compared, and the degree of coincidence between the binary image and the feature data is calculated as a statistical distance in accordance with the SRI algorithm in the same way as in step S2. This statistical distance calculation is performed for all 2 (U signal SC) temporarily stored in the video RAM 30.
This is performed by comparing each binary image with the feature data of the object related to the color data, but in this case, since the type of the part 10 is identified based on the color and shape, the identification in step S2 is performed. It has high accuracy compared to .

Thereafter, in step S7, the same identification result as in step S3 is made, and if the type of component 10 is identified as one, that is, the statistical distance of the first place is smaller than the certain value. , and the deviation from the second and third statistical distances is larger than the certain deviation value, and it is determined that the component 10 is the object for which the first statistical distance was calculated. At this point, a series of signal processing related to type identification is completed. In addition, if the statistical distance of the first place is larger than the above-mentioned 3 4 constant value, or if the deviation from the statistical distance of the second or third place is smaller than the above-mentioned certain deviation value even if it is smaller than the certain value, In step S8, it is determined that the signal cannot be identified, and the series of signal processing ends.

In this embodiment, steps SL, S2. and S3 correspond to the first identification step, steps 34, 35, S6 .
and S7 correspond to the second identification step.

Returning to FIG. 1, the CPU 18 also includes an external device I
/F36 is connected via the data bus line 28, and signals representing the above-mentioned type identification determination results, position measurement results, etc. are transmitted from the external device I/F36 to the parallel I/F3.
8 and serial number /F40 to the aforementioned sorting device, automatic processing machine, etc. For example, in a sorting device, the part 10 is sent to a sorting position according to its type, while in an automatic processing machine, the part 10 is operated to perform a predetermined processing depending on the position and type of the part 10. I am made to do so.

Here, when identifying the type of the component 10, the visual device of this embodiment first performs identification processing using a two-color image based on the luminance signal SY, and then uses color signals SR, SG, and S as needed.
Since a two-color image based on B is used, the color signal S
Compared to signal processing using R, SG, and SB,
The signal processing time is reduced while maintaining high identification accuracy. In particular, the identification using the color signals SR, SG, and SB is one of the top two or three in the identification based on the luminance signal SY.
Since the processing is performed only on different types of objects, there is an advantage that the processing time is further shortened.

Also, a luminance signal SY and color signals SR, SG.

When identifying the type of component 10 using SB, the signals are processed and binarized as analog signals, so it is difficult to identify them using gray scale images or color images based on digital signals. The amount of data to be processed is extremely small compared to the case of judgment, the signal processing time is significantly shortened, and a large capacity frame memory etc. is not required, so the device can be configured at low cost. .

5 6 In addition, in this way, the luminance signal SY and the color signals SR, SG
, SB are all performed based on two-color images, so the hardware and software for performing the identification process on two-color images can be shared. For example, when performing steps S6 and S7, steps S2 and S
There is an advantage that the device can be configured more simply and at a lower cost, such as by being able to use 3 programs.

Although one embodiment of the present invention has been described above in detail based on the drawings, the present invention can also be implemented in other embodiments.

For example, in the above embodiment, the color signal S of the three primary colors red, green, and blue is
R, SG, and SB are extracted and signal processed, but depending on the type of color of the object, only color signals related to two of these colors may be used, or color signals related to other hues may be used. It is also possible to

Further, in the embodiment described above, the color image detector 26 extracts two system numbers SC for a predetermined color, but it is also possible to have a microcomputer perform the same function.

Furthermore, in the above embodiment, the case where the type of object is identified by comparing the characteristic data of the object and the two-color image was explained, but normal two-color images of multiple types of objects are stored in advance, and in actual use. It goes without saying that other well-known identification methods may be used, such as a pattern matching method in which the two-color images obtained are matched and the degree of difference is determined. The method for determining the identification result is also changed as appropriate depending on the identification method and the like.

Furthermore, in the embodiment described above, identification using the color signals SR, SG, and SB is performed only for the top two or three types of objects identified using the luminance signal SY; however, the color signals SR, SG, and
All objects may be identified by SB.

Further, in the above embodiment, when the type of component 10 cannot be identified by identification using the luminance signal SY, the data signal SDR,
SD, , SD, are output to read the two system numbers SC regarding the color signals SR, SG, and SB, but the two system numbers SC regarding all objects are stored in the video RAM 34 in advance. Store the two plant numbers SC from the video RAM 34 as needed.
Alternatively, the identification process may be performed by reading out a binary image according to the method.

Further, in the embodiment, when the type of component 10 can be identified by identification using the luminance signal SY, the color signals SR, S
Although identification using G and SB is not performed, a confirmation process or the like may be provided in which a specified object is identified using color signals SR, SG, and SB for confirmation.

Although other examples are not provided, the present invention can be implemented with various modifications and improvements based on the knowledge of those skilled in the art.

[Brief explanation of the drawing]

FIG. 1 is a block diagram illustrating an example of a visual device in which object identification processing is performed according to the method of the present invention. FIG. 2 is a flowchart illustrating the operation of the apparatus shown in FIG. 1 when performing object identification processing. 10: 12: SP: R 3l. S4゜Part (object) Color camera (color imaging device) Color image signal SY: Brightness signal SG, SB: Color signal S2. S3: First identification step S5. S6. S7: Second identification step

Claims (1)

[Claims] Automatically identifies which of a plurality of predetermined types of objects an object imaged by a color imaging device is based on a color image signal output from the color imaging device. An object identification method, comprising: a first identification step of identifying the type of the object using a binary image obtained based on a luminance signal related to light intensity among the color image signals; and the first identification step. identifying the type of the object using a binary image regarding the color of the object obtained based on color signals regarding a plurality of hues among the color image signals; An object identification method comprising: a second identification step.