JPH02138677A - Detecting device for picture feature point - Google Patents

Abstract

PURPOSE:To simplify a process to detect the feature points of a 2-dimensional picture by using a filter means to detect the contours of a picture and the segment elements included in the contours with the same algorithm. CONSTITUTION:An arithmetic process part 4 includes a contour detecting part 61 which performs a filtering computing operation to detect the picture contours based on the digital picture signal. Furthermore a segment element detecting part 62 is added to perform a filtering computing operation to detect the seg ment elements included in the picture contours based on the detected contours and the given digital picture signal together with a feature point detecting part 7 which detects only the feature points of the contour lines based on the outputs of both parts 61 and 62. As a result, both the picture contours and the segment elements included in these contours can be detected via a filter means, i.e. with the same algorithm. Thus it is possible to simplify a process to detect the feature points of a 2-dimentional picture.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to an image feature point detection device that detects feature points in a two-dimensional image using image processing technology.

[Prior Art and Problems to be Solved by the Invention] Conventionally, as an example of a method for detecting features of a two-dimensional image, for example, a figure, a skeleton is converted into a point sequence, and each point is divided into 4 or 8 neighboring pixels. There is a method of detecting end points, intersection points, etc. based on information.

In this method, the curvature is also calculated by converting the skeleton into a series of points, and then calculating the angle formed by two line segments that connect each point to points on the skeleton that are separated by K (K is a positive integer) points before and after each point. Calculated based on change.

As described above, conventional methods require different algorithms depending on the type of feature points to be detected in order to detect feature points in an image, and the calculation time required for detection also increases. Ta.

The present invention was made to solve the above-mentioned problems, and aims to simplify the processing required to detect feature points in a two-dimensional image.

[Means for Solving the Problems] An image feature point detection device according to the present invention performs a first filtering operation for detecting the contour of an image, and performs a first filtering operation for detecting line segment elements included in the contour. The image forming apparatus includes filter means for performing a second filtering operation, and feature point detection means for detecting feature points of an image based on the results of the first and second filtering operations.

[Operation] In the image feature point detection device according to the present invention, the detection of the contour of the image and the detection of line segment elements included in the contour can be performed by the filter means, that is, by the same algorithm. As a result, the process for detecting feature points can be simplified.

[Embodiment of the Invention] FIG. 2 is a block diagram showing a hardware configuration of an image feature point detection apparatus showing an embodiment of the invention. Second
Referring to the figure, for example, a figure is imaged by a camera 1, and the image signal is given to an A/D converter 2. A/
The D converter 2 receives the image signal and provides the digitally modified image signal to the image feature point detection device 3. The image feature point detection device 3 includes an arithmetic processing unit 4 that performs arithmetic processing on digitally converted image signals, and a memory unit 5 that stores the processed data signals and necessary programs. As a result of the calculation, the calculation processing unit 4 outputs feature point information, shape information, etc. of the figure.

FIG. 1 is a block diagram showing an example of the arithmetic processing section 4 of the image feature point detection apparatus shown in FIG.

Referring to FIG. 1, the arithmetic processing unit 4 includes a contour detection unit 61 that performs a filtering operation to detect the contour of an image based on a given digital image signal, and a contour detection unit 61 that performs a filtering operation to detect the contour of an image based on a given digital image signal, and a A line segment element detection unit 62 performs a filtering operation to detect line segment elements in the contour based on the image signal, and only feature points on the contour are detected based on the outputs of the contour detection unit 61 and line segment element detection unit 62. and a feature point detection unit 7 for detection. Contour detection section 6
1 and the line segment element detection unit 62 both include digital filters having the following spatial characteristics f (x, y).

f (x, y) −x cloudy exp (−x2/2ax2−
>+2/2σy')/(2πσX'σy)... (1
) Here, σ8 and σ indicate deviations in the X direction and y direction, respectively.

FIG. 3 is an example of a characteristic diagram of a digital filter having the spatial characteristics shown in equation (1).

Next, filtering calculations in each of the contour detection section 61 and the line segment element detection section 62 will be explained.

In the contour detection unit 61, in order to obtain minute spatial characteristics with low direction selectivity in two-dimensional directions, the following equation (2) is given to σ8 and σ, and these values are converted to a length of several pixels. Set.

σ×1σy “+1 (2) This allows you to smooth a two-dimensional image to various degrees by changing the value of σ, and by performing a differential operation on the smoothed image, you can calculate the intensity change. It is possible to detect large parts of the image, that is, the outline of the image.

In the line segment element detection unit 62, in order to obtain spatial characteristics for detecting line segments with different curvatures, the following equation (3) is given to σ8 and σa.

σy−n·σX (3) Here, n is a positive integer. When n is changed, the length of the filter characteristic in the major axis direction changes.

If the threshold value is set to 1/2 of the maximum output of the filter, the detected curvature range changes by changing n. For example, by setting n large, the range of curvature to be detected becomes wider.

Next, referring to FIG. 1 again, the operation of the arithmetic processing section shown in FIG. 1 will be described.

First, the contour detection section 61 receives a digital image signal and processes it using a filter having spatial characteristics that satisfy the above-mentioned relationship of equation (2). Thereafter, empirically determined threshold processing is performed to extract the outline of the image. Next, the line segment element detection unit 62 applies a filter to the digital image signal having spatial characteristics that satisfy the relationship expressed by equation (3). As described above, since the range of detected curvature changes by changing n, the linearity of the contour obtained by the contour detection section 61 can be determined at various levels.

The feature point detection section 7 selectively outputs, from among the contours obtained by the contour detection section 61, portions where the output of the line segment element detection section 62 is equal to or less than a threshold value, that is, portions with low linearity. As a result, information regarding feature points of the image, such as end points, intersections, gaps, and high curvature parts of figures, can be obtained.

Note that the information about the contour obtained by the contour detection section 61 is used as shape information.

Note that the direction of the filter may be set, for example, in 36 directions every 10 degrees. Among the filters in 36 directions for each pixel, the filter that outputs the maximum value is defined as the direction of each pixel.

4 to 6 show output diagrams when the feature points of a specific figure are detected using the image feature point detection apparatus shown in FIG. 2. In both figures, (a
) indicates a figure given as an original image, and in the contour detection unit 61, σ8-〇, -2.0 . L threshold -2
, 5. The curves in (b) to d) of each figure indicate the contour of the original image obtained by the contour detection section 61, and the circles indicate the feature points obtained by the line segment element detection section 62 and the feature point detection section 7. show. (b) or d) in each figure
In these cases, the setting values in the line segment element detection section 62 are different.

In FIG. 4, (b) shows that the settings in the line segment element detection unit 62 are σX-2,0, σy-4,0°Threshold value-5
, 1, and (c) shows the case of σ, -2, 0, σ, -6
．． 0. The case of L threshold -7,6 is shown, (d) is σx
-2,0,σy-8°0, threshold value -10,1 is shown.

In Figures 5 and 6, (b) is σx-260, σ
2-5.0. The case of L threshold -6,3 is shown, and (c) is σ, -2,0, σy-6.0. The case of L threshold -7,56 is shown, and (d) is σ8-2. 0. σy−8
．． 0. The case of L threshold −10,0 is shown.

As can be seen from these output examples, it can be seen that the outline of the figure can be obtained, and at the same time, as the set value of σa increases, feature points can also be obtained for smaller changes in the outline.

In any of (b) to d), the end point portions, intersection portions, gap portions, and high curvature portions of the figure are obtained, and these are obtained as feature points.

The image feature point detection device shown in the above embodiment can be widely used as a character recognition or image recognition technique when automatically recognizing input documents, maps, photographs, etc., for example.

[Effects of the Invention] As described above, according to the present invention, the detection of the contour of an image and the detection of line segment elements included in the contour are performed by the same algorithm using the filter means, so that the characteristics of the two-dimensional image can be detected. An image feature point detection device with simplified processing for detecting points was obtained.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an example of an arithmetic processing section of the image feature point detection apparatus shown in FIG. 2, which is an embodiment of the present invention. FIG. 2 is a block diagram showing the hardware configuration of an image feature point detection apparatus showing an embodiment of the present invention. FIG. 3 is a characteristic diagram showing an example of the spatial characteristics of a digital filter used in the contour detecting section and line segment element detecting section shown in FIG. 4 to 6 show output diagrams when the image feature point detection apparatus shown in FIG. 2 is used for specific figures. In the figure, 1 is a camera, 2 is an A/D converter, 3 is an image feature point detection device, 4 is an arithmetic processing section, 5 is a memory section, 7 is a feature point detection section, 61 is an outline detection section, and 62 is a line This is a component detection section. Patent Applicant: A.T.R. Atsuyamaya Co., Ltd. (C) From the 2nd time onwards, 40 (b) (d) 50th lt, at f#. (b) (C) (d) No. 6 (a) Seat bird)#,. (b) (C) (d)

Claims (1)

[Scope of Claims] An image feature point detection device for detecting feature points indicating end points, intersections, gaps, or parts of large curvature of an image based on a two-dimensional digital image data signal, comprising: an image contour; filtering means for performing a first filtering operation for detecting the contour and a second filtering operation for detecting line segment elements included in the contour, based on the results of the first and second filtering operations; An image feature point detection device, comprising: feature point detection means for detecting feature points of an image.