JPH06195406A - Picture processing method - Google Patents

Abstract

PURPOSE:To improve the recognizing rate of line kind recognition by proceeding line segment trace at a branch point in a line kind recognizing technique in a drawing input device. CONSTITUTION:In a line segment tracing processing, a system sets a reference point to select a connection candidate vector (S1 sand 2). Even when the reference point is a branch point, if there is a proper vector, it is made to be the connection candidate vector. When an exceptional processing is not necessary (S3:No), a reference vector is moved to go to the next tracing (S4). When the reference point is the branch point and the reference vector and the connection candidate vector are connected (S3:Yes), the vector lengths of the reference vector is added to the connection candidate vector (S5). When the sum is not more than an upper limit threshold value (S6:Yes), the tracing goes to a next one. When the sum is over the upper limit threshold value (S6:No), the reference vector and the connection candidate vector are invalidated (S7). The tracing is finished (S8).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for recognizing characters, figures, etc., by performing line approximation (vectorization) on an image such as a hand-drawn drawing, and in particular, improving the line segment tracking function to improve line type recognition accuracy. For those that have improved.

[0002]

2. Description of the Related Art A hardware configuration of a drawing automatic input device is shown in FIG. The image scanner 1 optically reads a handwritten drawing or the like. The binary image input interface 2 takes in binary image data from the image scanner 1. The image memory 3 stores binary image data. The image processor 4 converts binary image data into vector data. A disk 5 is controlled by the disk controller 6. Reference numeral 7 is a system memory, 8 is a host system connection interface for communicating with a host system such as a CAD (Computer Aided Design) system and a mapping system, 9 is a host processor that performs the main control of this system, and each system unit is It is connected to the host processor 9 via the system bus 10.

An outline of the operation of this device is shown in FIG. The binary image input interface 2 takes in the binary image data read by the image scanner 1 and transfers / stores it in the image memory 3 (S1). After that, the image processing processor 4 performs the image processing on the binary image data by the thinning method or the contour detection / skeletonization method. In the case of the contour detection and skeletonization method, a contour pixel string is extracted from the binary image data to generate a contour vector (S2), and further, a skeleton vector corresponding to the center line of the contour vector is generated (S).
3). The contour vector data and the core line vector data are stored in the system memory 7. The host processor 9 separates and extracts vector elements such as character candidates, symbol candidates, and line segment candidates from these vector data (S
4). A character dictionary, a symbol dictionary, and the like are prepared in the system memory 7, and the host processor 9 performs pattern recognition with reference to the dictionary as necessary for various separated vector elements (S5 to 7). After that, the recognized pattern data is edited into a predetermined format as a recognition result (S8), and various output processings such as passing the recognition result to the CAD system are performed (S9).

Line segment processing (S
In 7), the algorithm shown in FIG. 11 is adopted in order to make it possible to recognize line segments including discontinuous parts such as broken lines and chain lines in addition to solid lines. In this algorithm, first, a line segment vector element that can be a candidate for a broken line component is extracted (S1). In this processing, the line segment vector elements are discriminated on condition that the length of the line segment vector elements is equal to or less than a threshold value (maximum line segment length set value). As a result, line segment elements corresponding to the solid line components are eliminated. Next, a search is performed for the extracted vector element group (S2), and an element that can serve as a reference line is selected (S3). After that, line segments are traced to group continuous elements (S4). Then, line type judgment is performed on the grouped elements (S5). In this line type determination, the distribution of the length of each element and the distance between each element is evaluated, and a line type such as a broken line or a chain line (further, a one-dot chain line or a two-dot chain line) is determined.

[0005]

In the conventional line type recognition,
In the line segment tracking, when the branch point of the line segment element was reached, the tracking of the line segment was ended.
Especially in the case of facility drawings and mechanical drawings, where there are many intersections of line segments and there are cases where solid lines and other line types are combined and described.
There is a problem that the tracking section is shortened and the number of cut-out points of the tracking section is increased, which becomes an obstacle to the line type recognition accuracy.

In view of such a problem, the present invention provides an image processing method for recognizing a line segment element and other vector elements, which enables continuous line segment tracking at a branch point and improves line type recognition accuracy. The purpose is to provide.

[0007]

Means and Actions for Solving the Problems
In order to achieve such an object, the line segment elements obtained by separating and extracting from the vectorized image are used as input data, the line segment elements are discriminated with the upper threshold for discriminating broken line elements, and the discriminated line segment elements are discriminated. In the method of tracking a line segment of a plurality of line segment elements and recognizing the line type of the detected line segment, the following procedure is added to trace the line segment element.

That is, when the branch point is reached, the tracing line element is not immediately disabled (terminated), but a tracing line segment element is selected from among the branching line segment elements according to a predetermined line segment selection algorithm to continue the tracing. It is possible. However, in order to eliminate the possibility of tracing the solid line portion having a branch point, the following processing is performed before continuing the tracing.

First, the combined line segment length of the pending line segment element and the tracking line segment element is calculated by an appropriate method. Then, the obtained combined line segment length is compared with the upper limit threshold value, and when the combined line segment length exceeds the upper limit threshold value, the line segment element combining the pending line segment element and the tracking line segment element is a solid line element. Is determined to be
Tracking ends after both line segment elements are invalidated. on the other hand,
When the combined line segment length is less than the upper limit threshold, it is determined that the line segment element that combines both line segment elements is a broken line element, and tracking is continued with the tracking line segment element as the traveling direction.

Similarly, in the line type recognition, the line segment elements connected via the branch points are recognized by imagining a composite line segment element and using it instead of the line segment element. Accuracy improvement can be expected.

[0011]

Embodiments of the present invention will be described below. The drawing automatic input device according to this embodiment is characterized by a broken-line broken line type recognition process, particularly a broken line tracking process. This processing function can be realized by software. In this case,
The hardware configuration may be an existing configuration as shown in FIG. 9, for example.

An outline of the line segment tracking process according to this embodiment is shown in FIG. As shown in this figure, first, an end point on the traveling direction side of a given reference vector is set as a reference point (S
1). Then, based on this reference point, another vector that is a connection candidate is selected (S2). In this processing, when the reference point or the end point of the connection candidate vector is the branch point, it is immediately determined whether or not the trace is possible without determining that the trace is impossible.

This connection candidate selection process will be described with reference to FIGS. As shown in FIG. 2, patterns 1 to 5 are assumed as patterns of line segment intersections, and patterns are divided by the algorithm shown in FIGS. In this algorithm, first, it is confirmed whether the reference point is a non-branch point or a branch point (S
1). If the reference point is a non-branching point, the process proceeds to the determination of pattern 1 (S2-5). That is, a non-branch point is searched within a predetermined area with the reference point as a base point (S2). Whether or not the connection is possible is determined based on the distance / angle between the end points, the angle with respect to the reference vector, etc. for the vector connected to the retrieved non-branch point (S4). As a result, when a vector that can be connected is found (S5: Yes), the vector is fixed as a connected vector. If no non-branch point is detected (S3: No) or a vector that can be connected is not found (S5: No), the process proceeds to the determination of pattern 2 (S6 to 9). That is, first, a branch point is searched based on the reference point (S6), and it is determined whether or not each vector connected to the branch point can be connected (S8). As a result, if a connectable vector is found (S9: Yes), the process proceeds to the next process. If no branch point is detected (S7: No), or if a connectable vector is not found even if a branch point is found (S9: No), it is determined that tracking is impossible (S10).

On the other hand, when the reference point is a branch point (S1:
No) proceeds to the determination of pattern 3 (S11, 12).
That is, first, it is determined whether each branch vector connected to the reference point can be connected (S11), and the connectable vector is determined as the connected vector (S12: Yes). If no connectable vector is found (S12: N
o), the process proceeds to the determination of pattern 4 (S13 to 16). That is, it is confirmed whether or not a connection error vector (see FIG. 2D) exists in the branch vectors connected to the reference point (S13), and the branch vector connected to the tracking destination side of the connection error vector candidate. Whether or not the connection is possible is determined for (S15). As a result, a vector that can be connected is determined as a connected vector (S16). When the connection error vector candidate is not found (S14: No) or when the connectable vector is not found (S16: No), the process proceeds to the determination of pattern 5 (S17 to S20). That is,
The non-branch point is searched by the same procedure as S2 (S17),
Whether or not the vector connected to the searched non-branch point can be connected is determined (S19). As a result, if the vector can be connected, it is determined as a connected vector (S20: Yes), and if the non-branch point is not found (S18: No) or the vector is not connectable even if found (S20: No), it is impossible to trace. (S10).

Returning to FIG. 1, after the concatenation candidate vector is selected by the above algorithm, it is confirmed whether or not exception processing is necessary (S3). In this process, it is determined whether the reference point is a branch point and the connected candidate vector is directly connected to the branch point. If this condition is not satisfied (S3: No), the reference vector is moved to the connection candidate vector and tracking is continued (S4). As a result, a broken line without an intersection is traced as shown in FIG. 5, for example.

On the other hand, if the above conditions are satisfied (S3:
Yes) proceeds to exception processing (S5, 6). First, the combined line segment length x is obtained by a method such as obtaining the sum of the line segment lengths of the reference vector and the connection candidate vector (S5). At this time, if the previous reference point is a branch point and the reference vector is connected to the previous reference point, the sum of the previous reference vector, the current reference vector, and the concatenation candidate vector is obtained, and this sum is combined. The line segment length is x. In this way, it is considered that there is no practical problem if consideration is given to whether or not the previous reference vector is added, but if necessary, a method of adding the reference vector two times before or before the reference vector can be adopted.

After this, the composite line segment length x is compared with the upper threshold value ln _max (S6). As a result, when x ≦ ln _max (S6: No), the reference vector and the connection candidate vector are invalidated, and the tracking is ended (S7, 8).
If x> ln _max (S6: Yes), the tracking is continued (S4). As shown in FIG. 6, in the processing of the reference vector x ₃ , x ₃ + x ₄ ≦ ln _max, and thus the tracking is continued. In the processing of the reference vector x ₆ , x ₆ + x ₇ > ln _max , so x ₆ , Invalidate x ₇ and end the trace with vector x ₅ . As a result, as shown in FIG. 7, even if the broken line intersects with another line segment, the line segment can be traced beyond the intersection and the line type can be accurately recognized. In particular, as shown in FIG. Even if the broken line portion and the solid line portion are continuous, the solid line portion and the broken line portion can be accurately distinguished.

[0018]

As described above, according to the present invention,
It has the following effects.

(1) Since the tracing direction can be appropriately selected at the branch point, even a broken line having a branch point can be traced before and after the branch point, and the line type recognition accuracy becomes longer as the tracing section becomes longer. Is improved.

(2) When the line segment elements to be traced are continuous via a branch point, the composite line segment length of the continuous line segment elements is obtained, and it is determined whether or not this composite line segment length is valid as a broken line element. Since the confirmation is performed, it is possible to eliminate the possibility that a solid line having a branch point is erroneously recognized as a broken line.

(3) Due to these advantages, the accuracy of line type discrimination is improved.

[Brief description of drawings]

FIG. 1 is a flowchart showing an outline of line segment tracking processing according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an assumed connection relationship pattern and a vector connection relationship corresponding to each pattern.

FIG. 3 is a flowchart showing a connected candidate vector selection process according to an embodiment of the present invention.

FIG. 4 is a flowchart showing a connected candidate vector selection process according to an embodiment of the present invention.

FIG. 5 is a diagram showing an example of an image to be processed.

FIG. 6 is a diagram showing an example of line segment elements to be processed.

FIG. 7 is a diagram showing an example of an image to be processed.

FIG. 8 is a diagram showing an example of an image to be processed.

FIG. 9 is a block diagram showing a hardware configuration of a drawing input device.

FIG. 10 is a flowchart showing an outline of the operation of the drawing input device.

FIG. 11 is a flowchart showing an outline of a line segment element recognition procedure.

[Explanation of symbols]

 1 ... Image Scanner 2 ... Binary Image Input Interface 3 ... Image Memory 4 ... Image Processor 5 ... Disk 6 ... Disk Controller 7 ... System Memory 8 ... Host System Connection Interface 9 ... Host Processor CPU 10 ... System Bus

Claims (1)

[Claims] 1. A line segment element obtained by separating and extracting from a vectorized image is used as input data, a line segment element is discriminated with an upper threshold for discriminating broken line elements, and the discriminated line segment element is traced. In the method of recognizing the line type of the detected line segment by detecting the line segment of the section that spans multiple line segment elements, when tracing the line segment element, when reaching the branch point,
A tracking line segment element is selected from branching line segment elements according to a predetermined line segment selection algorithm, a combined line segment length of the pending line segment element and the tracking line segment element is calculated, and is compared with the upper limit threshold value. If the segment length exceeds the upper threshold, the pending line segment element and the trace line element are disabled and the tracking ends, while if the combined line segment length falls below the upper threshold, the trace line element is An image processing method characterized by continuing tracking as a traveling direction.