JPH0624031B2 - Recognition device - Google Patents

Description

The present invention relates to a recognition device, and more particularly to a recognition device having a plurality of detection means, for example, a sensor and a plurality of processing devices, which has high reliability.

The conventional recognition device scans an object to be recognized with a light-receiving sensor or the like, collects all the data in one electronic computer, performs software processing, and recognizes the object from the relation between the data. . However, in this device, a large-capacity and high-speed large-scale computer is required to perform the data processing, and when a failure of the light-receiving sensor occurs, the failure location is detected and the recognition method is changed. There is a problem that a complicated algorithm is required, and when a failure occurs in the computer main body, the object recognition becomes completely impossible.

The present invention has been made in view of the above circumstances, and an object thereof is to solve the above problems in the conventional object recognition device, enable high-speed object recognition by a simple algorithm, and partly Another object of the present invention is to provide a recognition device capable of coping with a failure of the sensor or the like.

The above object of the present invention is to provide a plurality of input means for detecting a part of a recognition target and outputting corresponding data, and data output from a plurality of input means having a predetermined correspondence relationship among the plurality of input means. Identification based on data output from a first recognition processing device group consisting of a plurality of processing devices that calculate the degree of identification based on A second recognition processing means group consisting of a plurality of processing devices for calculating the degree, and an output means for outputting a recognition result for the recognition target based on the identification degree data from the first and second recognition processing apparatus groups. Has and
A failure is detected when data is not sent from the processing device to the output means, and the recognition result is obtained by using any one of the identification degree data obtained according to a processing method that can be changed according to the failure situation. It is achieved by a recognition device characterized by outputting.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of an object recognition device using an optical sensor according to an embodiment of the present invention. In the drawing, 1 to 5 are optical sensors, 10 to 50 are interfaces, 100 to 500 are recognition processing devices, and 10a. , 10b, 10c; 20a, 20
b, 20c; etc. are bus transmission lines, 1000 is an image display device, and 1000a is a bus transmission line. Also, the optical sensor 1
.About.5 and interfaces 10 to 50 are referred to as optical sensor devices.

The optical sensor devices 1 ′ to 5 ′ are bus transmission lines 10 a,
The recognition processing devices 100 to 500 (which are referred to as “level 1 recognition processing devices”) are connected in a mesh pattern through 10b, 10c; 20a, 20b, 20c; Is configured to transmit the processing result to the image display apparatus 1000 via the bus transmission path 1000a.

FIG. 2 shows an example in which the functions of the apparatus shown in FIG. 1 are expanded to realize different recognition methods and to improve reliability. In FIG. 2, symbols 1 to 5, 1'to 5 ',
10-50, 100-500, 1000 and 10a,
10b, 10c; 20a, 20b, 20c, etc., 1000
a indicates the same component as shown in FIG.
Reference numeral 500 'denotes the level 1 recognition processing device 100-500.
This is a level 2 recognition processing device, which is a lower level of the above.

FIG. 3 shows the interfaces 10 of the optical sensors 1'-3 '.
3 shows the connection status between the 30 and the level 1 recognition processing apparatus 200 and the level 2 recognition processing apparatuses 100 'to 300'. The interfaces 10, 20, 30 are connected to the level 1 recognition processing device 200 through the bus transmission lines 10c, 20b, 30a, and are further connected to the level 2 recognition processing devices 300 ', 200', 100 '. There is.
As a result, the recognition signal detected by the optical sensor device 2'is simultaneously transmitted from the interface 20 to the level 1 recognition processing device 200 and the level 2 recognition processing device 200 'via the bus transmission path 20b. The recognition processing device 200 of level 1 is a three-direction optical sensor device 1 ',
The object is identified based on the recognition signals from 2'and 3 '.

FIG. 4 shows the internal structure of the recognition processing device 200. The processing device 202 opens / closes the interface 201 at a constant cycle based on the timer 203 for the data sent from the bus transmission lines 10c, 20b, 30a. Are sequentially taken in by and are stored in the reception buffers 204, 205, 206 corresponding to the respective transmission lines 10c, 20b, 30a. next,
The processing device 202 performs processing according to a recognition algorithm described later based on these stored data, and stores the result in the processing result buffer 207. Further, the processing device 202 transmits the data in the processing result buffer 207 to the image display device 1000 via the interface 208. These series of processes are sequentially executed based on the timer 203.

FIG. 5 shows the internal structure of the image display device 1000. The processing device 1001 includes bus transmission lines 1000a and 10a.
The data sent from 00b is fetched by opening and closing the interface 1001 based on the timer 1003, and stored in the reception buffers 1004 and 1005 corresponding to the transmission lines 1000a and 1000b. Next, the processing device 1002 displays an image of the object on the display device 1006 from the data in the reception buffers 1004 and 1005 according to the display algorithm.

The process from object recognition to display will be described in detail below.

Now, it is assumed that the object S to be recognized is as shown in FIG. 6 (A), and the change in density of the object is shown by the curve D in FIG. 6 (B). When the optical sensor devices 1'to 6'are installed corresponding to the object S to be recognized, the respective optical sensor devices 1'to 6'receive light from the object S to be recognized and are shown in FIG. 6 (C). as,
Recognition values 0, 0, 0.5, 1, 1, 1 corresponding to the density
Indicates. These recognition values are the interface values 10 to 10.
A recognition signal is transmitted via 60 to the recognition processing device.

The recognition processing device 200 of level 1 includes an optical sensor device 1 ′,
The recognition signals from 2'and 3'are transferred to the bus transmission lines 10c and 20.
b, 30a. Further, the level 2 recognition processing device 300 'sends the recognition signals from the optical sensor devices 1', 3 ', 5'to the bus transmission lines 10c, 20b, 30.
Receive via a. That is, the level-1 recognition processing device receives the recognition signals from the directly corresponding positions and the photosensor devices on both sides thereof, and the level-2 recognition processing device receives the recognition signals on the directly corresponding position and two adjacent photosensors on both sides thereof. Receive a recognition signal from the device. Each recognition processing device calculates the identification degree P _{j of the} object based on these three recognition signals as described below.

P _j = k ₁ + v _i + k ₂ v _j + k ₃ v _k (1) where v _i + v _j and v _k are recognition values by the optical sensor device, and v 1 in the level 1 recognition processing device. _i = v _j−1 , v _k = v _{j + 1 In the} level 2 recognition processing apparatus, v _i = v _j−2 , v _k = v _{j + 2} , and k ₁ , k ₂ , k ₃ are defined. Is a constant. In the formula (1), in particular the constant _{k 1,} k 2, _{k 3} and _{k 1 = -0.5, k 2 =} 1, k 3 = -0.5 Recognition method of suppressing scheme case of a k ₁ = - The case where 1, k ₂ = 1 and k ₃ = 0 is referred to as a differential recognition method, but the method of deciding the constants k ₁ , k ₂ and k ₃ is not limited to this.

Hereinafter, a concrete example of the recognition method of the suppression method will be described with reference to FIG.
Description will be made based on (D).

Recognition signals 0, 0, from the above-mentioned optical sensor devices 1'-6 '
Each of the level 1 recognition processing devices 100 to 600 that received 0.5, 1, 1, 1 receives the equation P _j = −0.5v _j−1 + v _j −0.5v _{j + 1.}
The discriminating degree is calculated by (2), and 0, -0.25, and
We get 0,0.25,0,0. Further, each of the recognition processing devices 100 ′ to 600 ′ of level 2 has the formula P _j = −0.5v _j−2 + v _j −0.5v _{j + 2} ...
The discrimination degree is calculated according to (3), and the discrimination degrees are −0.25, −0.
5, -0,0.5,0.25,0 are obtained.

This is graphed (D), where the graph shows the degree of recognition by the level 1 recognition processing device, and the graph shows the degree of recognition by the level 2 recognition processing device. From these graphs, the recognition target S
It can be seen that the outline of is highlighted. Also, level 1
Comparing the recognition degrees of the recognition processing device of level 2 and the recognition level of the recognition processing device of level 2, the contour position of the recognition processing device of level 2 is slightly blurred.

Next, how to deal with the case where a failure occurs in the optical sensor device will be described with reference to FIG.

Now, assuming that the optical sensor device 3'has failed, the recognition processing device 20 in which the input from the optical sensor device 3'is lost.
For 0, 300, 400, the discrimination degree is calculated based only on the input from the other optical sensor device, and 0, -0.5, 0.5 is obtained (see FIG. 7 (A)). Further, assuming that two consecutive photosensor devices 3 ', 4'fail, the recognition processing devices 200, 300, 400, 500 calculate the discrimination degrees as 0, 0, -0.5, 0.5, respectively. (See FIG. 7 (B)). Graphs of these are shown in Fig. 7 (C), and the graph shows the case where there is no failure in the optical sensor device (Fig. 6 (D)).
The same as the graph of FIG. 2), the graph shows the case where the optical sensor device 3 ′ fails, and the graph shows the optical sensor device 3 ′,
The case where 4'has failed is shown respectively. As is clear from the graph, even if a failure occurs in the optical sensor device, the position of the recognized object S is not overlooked, but the object can be recognized, although some positional error occurs. It should be noted that the recognition processing device, which has detected that there is no input from the optical sensor device and is out of order, displays an abnormality of the optical sensor device on the display device.

Next, how to deal with a failure in the recognition processing device will be described with reference to FIG.

Now, assuming that the level 1 recognition processing devices 200 and 400 have failed, the image display device 1000 is recognized by the recognition processing device 2.
Since the identification degree data is not sent from 00 and 400, these failures are detected and displayed on the display device 1006. The first method of correspondence is a method of using the identification degree of the level 2 recognition processing apparatuses 100 'to 600', and the second method is a failed level 1 recognition processing apparatus 200, 400.
This is a method of using, instead of the identification degree of each of the above, the identification degree of the corresponding level 2 recognition processing devices 200 'and 400'.
The graph of FIG. 8 (B) shows the degree of discrimination by the second method, and the recognition processing devices 200, 40 of level 1
A graph of the degree of discrimination when 0 is not a failure (Fig. 6
Compared with (the same as the graph of (D)), it can be seen that the object can be recognized without overlooking the outline of the recognized object S.

The image display device 1000 is not limited to the above-described abnormal time,
It goes without saying that various identification methods can be arbitrarily selected even in the normal state. Further, in the above-mentioned embodiment, an example in which the optical sensor devices are arranged one-dimensionally is shown, but it goes without saying that a multi-dimensional object can be recognized by arranging the optical sensor devices in a multi-dimensional manner. Further, in the above-described embodiment, the object is recognized by the optical sensor device, but this is not limited to light (visible light), and light other than visible light (infrared rays, ultraviolet rays, etc.), sound waves, and other objects can be recognized. It goes without saying that a medium may be used.

As described above, according to the present invention, in an object recognition device using a plurality of sensors, a plurality of processing devices are provided, and the plurality of sensors and the plurality of processing devices are connected in a mesh shape, thereby making Each of the sensors recognizes only a part of the object to be recognized and sends this signal to the plurality of processing devices, and each of the plurality of processing devices receives a signal based on only the recognition signals from the plurality of adjacent sensors. Since it is configured to recognize changes in the image of the object to be recognized, it is possible to recognize the object at high speed with a simple algorithm, and it is also possible to deal with failure of some sensors, etc. The remarkable effect that a low-priced object recognition device can be realized is achieved.

BRIEF DESCRIPTION OF THE DRAWINGS FIGS. 1 and 2 are configuration diagrams showing an embodiment of the present invention, FIGS. 3 to 5 are configuration diagrams showing details of each part, and FIGS. 6 to 8 are books. It is a figure explaining operation of an invention. 1 to 5: optical sensor, 1'to 6 ': optical sensor device, 10
50: Interfaces, 10a, 10b, 10c; 2
0a, 20b, 20c; ...: Bus transmission path, 100 to 6
00: Level 1 recognition processing device, 100 'to 600':
Level 2 recognition processing device, 1000: image display device.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Koji Miyamoto 1099, Ozenji, Tama-ku, Kawasaki City, Kanagawa Prefecture, Ltd., within the Systems Development Laboratory, Hitachi, Ltd. Hitachi, Ltd. System Development Laboratory (72) Inventor Hirokazu Ihara, 1099, Ozenji Temple, Tama-ku, Kawasaki City, Kanagawa Stock Company, Hitachi Systems Development Laboratory (56) Reference JP-A-57-155672 (JP, A) JP-A-55-124878 (JP, A)

Claims (1)

[Claims] 1. A plurality of input means for detecting a part of a recognition target and outputting corresponding data, and data output from a plurality of input means having a predetermined correspondence relationship among the plurality of input means. Based on data output from a first recognition processing device group including a plurality of processing devices for calculating the identification degree and a plurality of input means having a correspondence relationship different from the predetermined correspondence relationship, A second recognition processing means group composed of a plurality of processing devices for calculation, and an output means for outputting a recognition result for the recognition target based on the identification degree data from the first and second recognition processing device groups. A failure is detected when data is not sent from the processing device to the output means, and the identification level data obtained using a processing method that can be changed according to the failure condition is used. Recognition result Recognition apparatus and outputting.