JPH03255302A - Detecting apparatus for pattern - Google Patents

Abstract

PURPOSE:To detect the abnormality of a body to be inspected automatically and to reduce the capacity of memories by projecting the picture-element components of an input-image signal in both X and Y directions, counting the components, obtaining the characteristic amounts, and detecting and judging the abnormality of the body to be detected. CONSTITUTION:A moving robot 1 is moved to a specified position. The image of an insulator 3 is picked up with a camera 2. The binary-coded data are inputted into an image input means 4 of a pattern detecting apparatus 20. Differentiation is performed in a directive differentiating circuit 5. The concentration gradient of the insulator 3 in the direction of the water drip shed is detected, and the direction components are obtained. The results of the differentiations in the upper and lower directions are expressed as the directional differentiated signals in the directions of X, Y and Z. With respect to the picture elements wherein the absolute values of the differentiated values are higher than a specified threshold value, the numbers of the picture elements in the direction perpendicular to the water drip shed are counted in counting means 6 separately for the positive and negative values. Then a projection step is performed. The region whose number of the picture elements is larger than the specified threshold value is detected as the candidate region for the object of the insulator 3. The width, height, central position and constituent picture elements of the insulator region are computed based on the projection data in a processor 9. The characteristic amounts are compared with reference values stored in a memory means 8 in an abnormality detecting means 7. Thus the intended insulator 3 can be detected.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a pattern detection device for moving a camera mounted on a mobile robot to a predetermined location such as a substation and performing a tour inspection of equipment. be.

[Prior art] Figures 9 (a) and (b) are, for example, published by Shokodo, 1986 6.
“Image Processing Handbook” (Page 613-
FIG. 615) is an explanatory diagram of a binary image pattern and a partial pattern by a recognition method called a conventional composite partial pattern matching method shown in FIG. This pattern detection method is
A characteristic partial pattern P+ among the target binary patterns.

Select p, , p, . . . and store them in the computer in advance. When the object comes within the field of view, the computer compares the partial pattern with the standard pattern and determines the positional coordinates of the point on the screen that best matches. The PI obtained in this way,
The distance d+□ and angle θ1□ calculated from the coordinates of P2 are
If it matches the geometrical relationship of the target object, it is determined that both P+ and P2 have been calculated correctly.

[Problem to be solved by the invention]

Conventional pattern pressure detection devices are configured as described above, so if the distance to the object changes or the size of the object changes in the image, the standard pattern and the inspection time change. There was a problem in that it was difficult to match the pattern with the partial pattern, and accurate positioning could not be achieved.

Furthermore, many standard patterns must be stored in the storage device as image data, which poses problems such as requiring a large capacity memory.

This invention was made to solve the above-mentioned problems, and can be easily applied even when the size of an object in an image changes, and the storage capacity for the object can be reduced. An object of the present invention is to obtain a pattern detection device that requires less.

[Means to solve the problem]

The pattern detection device according to the present invention includes an image input means for capturing an image of an object to be inspected with a visual device and quantizing the image signal, and a density gradient for obtaining a density gradient in a predetermined direction in each pixel of the image signal. detecting means, and X, Y detection means for pixels whose absolute value of the direction component of the density gradient is above a certain amount.
The difference between the information processing means that projects the pixels in the direction, counts the pixels using the counting means, and calculates the feature amount from the counted value, and the reference value that measures the feature amount in advance and stores it in the storage device is within an acceptable range. The apparatus is equipped with an abnormality detection means for determining whether or not the object is within the range of the object to be inspected.

[Effect]

The pattern detection device according to the present invention obtains a feature amount by projecting and counting pixel components having a density gradient in a certain direction in both the X and Y directions from an input image signal captured by a visual device, and compares it with a reference value. Since the abnormality detection judgment of the object to be inspected is made using the method, the feature amount of the object to be inspected is emphasized, and even if there is a change in the size of the object in the image, it is reliably handled.

[Embodiment 1 of the Invention Hereinafter, an embodiment of the present invention will be described with reference to the drawings. 1st
The figure is an explanatory diagram for explaining an actual situation in which the present invention is applied. In the figure, 1 is a mobile robot, 2 is a camera as a visual device, and 3 is an insulator of an object to be detected, for example. .

Next, the operation will be explained. First, when automatically performing maintenance and inspection of equipment at a substation, power plant, etc., the mobile robot 1 moves to a predetermined position at the inspection location and photographs the insulator 3 with the camera 2. When performing an external appearance inspection, only the target object is detected from the image by distinguishing between the insulator 3 and other structures.

FIG. 2 is a block diagram showing an embodiment of hardware for configuring this invention. In the figure, 4 is a camera 2.
5 is a directional differentiation circuit as a concentration gradient detection means for detecting a concentration gradient from an image of the object; 6 is a counting means; 7 is an abnormality for determining whether the object is normal or not. A detection means, 8 a storage means, and 9 a processor CPU serving as an information processing means.

First, the mobile robot 1 moves to a predetermined position. Then, the insulator 3 to be inspected is imaged by the camera 2, and digital image input data is taken into the image input means 4 of the pattern detection device 20 (step 5TI). FIG. 4 is an example of an input image, in which an input image containing background structures other than the insulator of the object to be detected appears. In order to detect the density gradient in the direction of the folds of the insulator 3 (from top to bottom in the figure) from the input image and extract the directional component, a directional differentiation circuit 5 with unidirectionality is used as shown in FIG. (Step 5T2). For example, when the folds are horizontal, Fig. 6 (C
) is used. This differential operator only needs to be able to detect positive and negative direction components of edges such as folds. In the case of FIG. 6, the hatching in (A) represents the image of the object. The results of differential processing in the upward and downward directions at the positions of Xi, Yl, and Zl are respectively shown in X2. Y2, Z
It is expressed as an electric signal of a directional differential circuit of 2. The signal level increases at the horizontal edges of the object, and the density gradient is represented by the sign of the signal.

Therefore, the directional differentiation circuit 5 only needs to be an operator capable of detecting positive and negative direction components (step 5T3). With respect to this differential image, the counting means 6 counts the number of pixels in the direction perpendicular to the folds, separately for positive and negative pixels, for pixels for which the absolute value of the differential value is greater than or equal to a predetermined threshold.

FIGS. 7(a) and 7(b) are projection data of the counted periodic patterns in the axial direction. For this projection data, an area where the pixel count value n is equal to or greater than a predetermined threshold value 78 is designated as a horizontal target object candidate area of the insulator 3, R,, R2, R,...
... is detected (steps ST4, 5T5).

Next, each of the object candidate regions R,,R,.

The number of pixels in the fold direction is counted for the inside of R3, and projection processing is performed as shown in FIG. 2 (step 5T6).
, an area where the number of pixels is equal to or greater than a predetermined threshold value of 78 is detected as an insulator object candidate area. Furthermore, from the two projection data shown in FIGS. 7 and 8, the radius W(R,
), the height H1 center position (x, y), and the constituent pixels S are calculated by the processor 9 (step 5T7). These feature quantities are measured in advance and compared with reference values stored in the storage means 8 by the abnormality detection means 7 to fix the pattern and detect the target insulator 3 (step 5T8).
).

In the comparison with the reference value, as shown in equation (1), if the difference between each feature amount and the reference value is within a predetermined acceptable range, it is assumed that the target insulator has been detected. Also, if it is outside the accepted range, it is determined to be abnormal and an abnormal signal is output (step 5T).
9).

However, WX1△W: Standard value of width W and acceptable range HK, △H
: Reference value of H and acceptable range SX, △S: Reference value of width S and acceptable range abc : Absolute value Also, as an evaluation value, the density of object occupancy in the insulator existence candidate area ρ=s/(WXH) Area aspect ratio W/
H may also be used to produce the same effect as in the above embodiment.

Furthermore, in this embodiment, the detection of the insulator was explained.
The same method can be applied to objects that have repeating components in the same direction, such as stairs, ladders, and electric wires.

〔Effect of the invention〕

As described above, according to the present invention, there is an image input means for taking in an image signal of a photographed object to be inspected, a density gradient detection means for calculating a density gradient in a certain direction from the image signal, and The apparatus is configured with an information processing means that calculates feature quantities by counting pixels by projecting in the Y direction, and an abnormality detection means that detects abnormalities in the object to be inspected by comparing the feature quantities with a reference value. It can be easily applied to changes in the size of patterns, and has the effect of expanding the applicable range of pattern detection and improving detection reliability. In addition, there is no need to have multiple standard patterns, and the memory capacity is small, resulting in a smaller size and lower cost.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram of an application example showing an embodiment of the present invention;
FIG. 2 is a block diagram showing a hardware configuration according to an embodiment of the present invention, FIG. 3 is a flowchart showing a procedure for detecting an object to be detected, FIG. 4 is an explanatory diagram of an input image, and FIG. The figure is an explanatory diagram of the directional differential image, and Figure 6 is an explanatory diagram of the differential operator and waveform for detecting horizontal edge components.
Figure 7 is a projection data characteristic diagram that counts the number of pixels in the vertical direction of the image in Figure 5, Figure 8 is an explanatory diagram of horizontal projection data and evaluation values for pattern detection, and Figure 9 is a conventional diagram. FIG. 2 is an explanatory diagram of a pattern detection method. In the figure, 2 is a camera (visual device), 4 is an image input means, 5 is a directional differentiation circuit (concentration gradient detection means), 6 is a counting means, 7 is an abnormality detection means, 8 is a storage means, and 9 is a processor ( information processing means). In addition, in the figures, the same reference numerals indicate the same or equivalent parts.

Claims (1)

[Claims] an image input means for capturing an image of an object to be inspected in which a unidirectional component is repeated by a visual device and quantizing the image signal; and a concentration gradient for determining a component in a predetermined direction of the concentration gradient at each pixel of the image signal. a detection means;
An information processing means that counts the pixels using a counting means that projects in the Y direction and calculates a feature quantity from the counted value, and a difference between the feature quantity and a reference value that has been measured in advance and stored in a storage means. A pattern detection device comprising an abnormality detection means for detecting whether or not it is within an acceptable range.