JPH0571989B2 - - Google Patents

Description

[Detailed Description of the Invention] [Technical Field of the Invention] The present invention relates to a binary image conversion device, and particularly relates to a binary image conversion device that converts a binary image containing a mixture of line figures and area figures. be.

[Technical background of the invention and its problems]

When file image information such as design drawings, maps, weather maps, etc. to a computer system or perform pattern recognition, this image information is binarized, and the resulting binary image is compressed and characterized. Perform extraction. However, this image information usually includes a mixture of solid area figures and line figures that originally do not have shape information. FIG. 2 shows an example in which area figures and line figures are mixed.

In order to efficiently extract line information, a method has conventionally been used in which an input binary image is converted into an image with a line width of 1, and this method is called image thinning processing. Regarding line thinning processing for binary images, see "Various considerations on line thinning methods" by Hideyuki Tamura.
(IEICE Technical Research Report, PRL Study Group Materials
PRL75-66, 1975), Japanese Unexamined Patent Publication No. 163078/1983, etc.

Normally, line thinning processing is performed by binarizing a target image and sequentially removing contour points of the obtained binary image. When this thinning process is applied to an image containing a mixture of area figures and line figures as shown in Figure 2, the original shape information of area figures is lost as shown in Figure 3. It gets lost.

On the other hand, when performing the contour tracking process described in Tominaga et al.'s ``Proposal of Two-dimensional Block Transmission by Boundary Line Tracing of Binary Figures'' (1971 Institute of Electronics and Communication Engineers National Conference No. 1330), Although the shape information of the surface figure is saved and extracted, the outline of the extra line figure part is also traced, which poses a problem in terms of processing time.

[Purpose of the invention]

This invention was made in order to solve the above-mentioned problems, and aims to enable data compression and feature extraction to be effectively performed on binary images in which line figures and area figures coexist. It is something.

[Summary of the invention]

This invention includes a thinning means for converting a binary image containing a mixture of line figures and area figures into a figure with a line width of 1;
By converting the value image using two processing systems, the area figure detection means, which detects the area figure part, and combining the processing results of both systems, only the line figure part has a line width of 1.
This is to obtain an image converted into a figure.

〔Effect of the invention〕

According to this invention, since it is possible to obtain a binary image in which area figures and line figures are distinguished, it is possible not only to perform data compression and feature extraction effectively, but also to use a simple circuit configuration. It becomes possible to perform conversion processing on binary images.

[Embodiments of the invention]

FIG. 1 is a diagram for explaining an embodiment of a binary image conversion device according to the present invention.

This binary image conversion device includes a thinning circuit 1, a surface figure detection circuit 2, and a compositing circuit 3. An image containing a mixture of area figures and line figures is read using a photoelectric conversion device such as a CCD sensor, and is converted into binary image data consisting of M×N pixels by binarization processing. f _ij indicates the value of the j-th line and i-th pixel in the M×N binary image, and each image data is sequentially input into the line thinning circuit 1 and the surface figure detection circuit 2.

The line-input binary image data is converted into a figure with a line width of 1 by the line thinning circuit 1, and the plane figure information is extracted by the plane figure detection circuit 2. Based on these output results, the synthesis circuit 3 encodes the image data C, which consists of an image in which only the line figure information is converted into a figure with a line width of 1, and an image obtained by extracting the surface figure information. _ij is output.

FIG. 4a is a diagram for explaining the thinning circuit 1 of FIG. 1 in detail.

The thinning circuit 1 is realized by performing n-stage pipeline processing using logical filtering consisting of a 3×3 mask (filtering table). Since the number of stages n of the pipeline 40 is determined by the width of the line figure, it is sufficient to assume an upper limit of the line width and set it to a size sufficient to thin the line figure of this line width.

FIG. 4b is a diagram for explaining in detail the circuits forming each stage of the pipeline 40.

The circuits forming each stage of the pipeline 40 include a line buffer 41, a shift register 42, an address register 43, and a filtering table 44. In this embodiment, ring filtering consisting of a 3 x 3 mask is used, so the input signal is input to a line buffer 41 consisting of three stages, and a 3 x 3 binary image signal is input to a shift register 42. be remembered. The binary image signals stored in the shift register 42 are sequentially read out by an address register 43 and processed by a 3×3 filtering table 44.

FIG. 5 is a diagram for explaining in detail the surface figure portion detection circuit 2 of FIG. 1.

The surface graphic portion detection circuit 2 is composed of a 2n-stage pipeline 50, and after shrinking by n pixels through the pipeline processing through the n stages, it expands by n pixels through the pipeline processing through the same n stages. A circuit similar to that shown in FIG. 4b is used for each stage of the pipeline. After shrinking in this manner, expansion is performed to smooth out convex portions of the binary image. That is, it is possible to erase the line figure part and detect the area figure part.

6a and 6b show the synthesis circuit 3 of FIG.
FIG. 2 is a diagram for explaining in detail.

This synthesis circuit 3 includes a line buffer 61, an address register 62, and a conversion table 63.
The output A of the thinning circuit 1 and the output B of the surface figure detection circuit 2 are input to this synthesis circuit 3 and both output data are synthesized. In order to adjust the timing of both output data, the output A Line buffer 61 on the side
will be established. The output data A and the output data B whose timings have been matched are sequentially read out by the address register 62, and are code-converted using the conversion table 63, and the background, line figure part, and surface figure part are distinguished and coded. The converted image data C _ij is output.

FIG. 6b shows the contents of the conversion table 63. According to this conversion table 63, the line figure information is “1”
In this case, the surface graphic information is encoded as "2".

In this conversion table 63, in the relationship between output A as a result of thinning and output B as a result of area figure extraction, when output A is "1" and output B is "1", that is, when the thinning process In the case where the surface figure is also thinned in the process, the composite output C _ij is set to be "2" so that the output B is prioritized.

FIG. 7 shows an example of conversion of a binary image when n=3. In the figure, a shows an input binary image, b shows an image as a result of pixel contraction, c shows an image as a result of pixel expansion, and d shows an image as a result of code conversion by a synthesis circuit.

As is clear from a comparison between the input binary image in Figure 7a and the converted image in Figure 7d, it is extremely easy to extract features from graphical information by performing image conversion according to the present invention. becomes. For example, end points of line figures such as C and F, or D,
It is easy to detect points of contact between line figures such as E and area figures. In addition, the shape information of a surface figure can be easily extracted by finding the outline of the part with the code "2", and furthermore, the connection relationship between the surface figures can be determined by tracing the pixels whose code is "1". This can be easily done by

[Other embodiments of the invention]

FIG. 8 is a diagram for explaining another embodiment of the binary image conversion device according to the present invention.

If the area figure appearing in the drawing is limited to a black circle with a constant size as shown in FIG. 8a, for example, the area figure part detection circuit 2 shown in FIG. 5 can be further simplified. An example of this is shown in FIG. 8b.

This surface figure detection circuit 2 includes a black pixel count circuit 81, a comparison circuit 82, and an m pixel expansion circuit 8.
3, a line buffer 84, and an AND circuit 85. The binary image data f _ij is input to a black pixel number counting circuit 81, and the number of black pixels existing within a 2m×2m window is counted. The counted result is compared with a threshold value T by a comparison circuit 82, and the center of the black circle is detected. That is, the threshold value T is set so that when a pixel is detected near the center of a black circle, the number of black pixels within a 2m x 2m window becomes greater than the threshold value T. Next, the m-pixel expansion circuit 83 expands the data by m pixels, and then the AND circuit 85 performs an AND operation with the input binary image data f _ij and deletes the redundant expanded portion. In addition, in order to adjust the timing, input binary pixel data is input via a line buffer 84.
Input to AND circuit 85.

Note that the binary image conversion device according to the present invention is not limited to the above embodiments. That is, for example, it is also possible to first input the entire binary image to the image memory and then access the image memory without using pipeline processing. In this case, the image memory is accessed, the processing using the thinning means and the surface figure detection means is performed individually, the results are stored in separate memories, and then both memories are accessed to perform the compositing process. All you have to do is Also, as a synthesis circuit, it can be used as a simple OR
There may be cases where a circuit is sufficient.

[Brief explanation of drawings]

FIG. 1 is a diagram for explaining an embodiment of a binary image conversion device according to the present invention, FIGS. 2 and 3 are diagrams for explaining the prior art, and FIG. 4 is a diagram showing thin lines in FIG. 1. A diagram to explain the circuit in detail,
FIG. 5 is a diagram for explaining in detail the surface figure detection circuit in FIG. 1, FIG. 6 is a diagram for explaining in detail the synthesis circuit in FIG. 1, and FIG. FIG. 8, which is a diagram showing an example of the result of image conversion processing, is a diagram for explaining another embodiment of the binary image conversion apparatus according to the present invention. In the figure, 1... line thinning circuit, 2... surface figure detection circuit, 3... synthesis circuit, 41, 61... line buffer, 42... shift register, 43,
62... Address register, 44... Filtering table, 63... Conversion table, 81...
Black pixel count circuit, 82... Comparison circuit, 83
... m pixel expansion circuit, 84 ... line buffer,
85...AND circuit.

Claims (1)

[Scope of Claims] 1. A binary image conversion device for processing a binary image containing a mixture of area figures and line figures, comprising: thinning means for thinning the binary image into figures with a line width of 1; A plane figure detection means for extracting only a plane figure part from a value image; based on the data of the figure thinned by the thinning means and the data of the plane figure extracted by the plane figure detection means. compositing means for encoding and compositing, using a conversion table during encoding in the compositing means, and using the conversion table, among the figures thinned by the thinning means, the area figure part detection means The data excluding the portion corresponding to the surface figure portion thus extracted and the data of the surface figure portion extracted by the surface figure portion detecting means are set to be encoded separately. A binary image conversion device. 2. A patent claim characterized in that the plane figure detection means is set to perform contraction of the binary image by n pixels so that only line figures can be erased, and then expand it by n pixels. Range of 2 described in item 1
Value image conversion device.