JPH06325176A - Image extracting device - Google Patents

Abstract

PURPOSE:To shorten the time required for an outline converging process after outline designation, to reduce the influence of manual outline designation, and to prevent convergence on a wrong outline. CONSTITUTION:The image extracting device displays a source image in a source image memory 40 on a display 46, specifies an outline by an outline specifying means, and converges the specified outline on a specific outline by an outline converging means, thereby extracting part of the source image in the converged outline. The outline specifying means detects the intersection of an input segment on a tablet 30 and the outline of the source image on the display 46, stores its intersection information in an intersection information memory 38, and also connects intersections corresponding to the intersection information to generate a polygonal outline, thereby outputting it as a specified outline to the outline converging means. Then the segment crossing the outline of the source image is only inputted on the tablet 30 to automatically determine the intersections forming the specified outline, so that the outline designation can be performed with high accuracy.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image extracting device used in an image simulation system, an electronic catalog, a CM production machine and the like. More specifically, an original image (or an original image) captured by using an image input unit (for example, a television camera) is stored in an original image memory and is displayed on an image display unit (for example, a CRT display), and is displayed by a contour designating unit. The present invention relates to an image extracting device which specifies a contour, converges the contour specified by the contour specifying means to a predetermined contour line, and extracts a part of an original image within the converged contour line. .

[0002]

2. Description of the Related Art Generally, an image extracting apparatus of this type mainly uses a system using a GWS (graphic workstation), and a part (specific image) of an original image (for example, a natural image) is semi-automatically operated. Object such as person or object)
The conventional image extracting apparatus shown in FIG.
And the means for realizing the functions shown in FIG.

That is, the original image 12 and the corresponding image data on the display screen 10 of the image display unit as shown in FIG. 6 (a) are moved from a tablet (or mouse) as an example of a pointing device. Based on the input, as shown in FIG. 6B, a rough contour line 16 representing the outline of the contour of the target object (for example, a person) 14 to be extracted from the original image 12 is designated (or outside the target object 14). Specify an outer contour line (not shown) that surrounds the.

Then, the image extracting device stores the outline information input from the tablet in the corresponding memory, and if the closed loop is completed, "Is closed loop completed?"
S ”, and the contour designation process ends. If the closed loop has not ended, “NO IN CLOSED LOOP?” Becomes “NO”, and outline contour designation from the tablet is repeated until “YES?” In “Ended closed loop?”.

When the contour specifying process is completed, the image extracting apparatus performs the thinning process (or the active contour model process (SNAKES)) of the rough contour line 16 by the contour convergence function of the system based on the automatic contour convergence program. 6C, it converges on a predetermined contour line 18 as shown in FIG. 6C, and the object 14 is extracted in this converged contour line 18 as shown in FIG. 6D.

[0006]

However, in the conventional image extracting device, the manual input from the tablet
A rough contour line 16 representing the outline of the contour of the object 14 to be extracted from the original image 12 is designated, or an outer contour (not shown) of the object 14 to be extracted from the original image 12 is designated to designate the contour. Since I was doing the following, (1), (2),
There was a problem as described in (3).

(1) If an attempt is made to shorten the input time of the rough contour line 16 by manual input using a tablet, the deviation between the center line of the rough contour line 16 and the contour line of the extraction object 14 becomes large, and the contour is designated. There is a problem that the time required for the subsequent contour convergence processing becomes long. (2) When the thinning processing of the rough contour line 16 is selected as the contour convergence processing, the center line of the rough contour line 16 and the extraction target 1
The larger the deviation from the contour line of No. 4, the more adversely affects the quality of the extracted image.

(3) When the dynamic contour model processing (SNAKES) is selected as the contour convergence processing, a part 18a of the converged contour line 18a is pulled by the background line 20 having a high brightness in contact with the outer contour line. Is easily converged to an erroneous contour line (originally, the contour line indicated by the chain double-dashed line 18t) as shown by the solid line in FIG. 6C, and the background line as shown in FIG. There is a problem that a part 20a of 20 is displayed.

The present invention has been made in view of the above problems, and it is possible to shorten the time required for the contour convergence processing after the designation of the rough contour, and to reduce the influence of the contour designation by manual input. It is an object of the present invention to provide an image extraction device capable of preventing convergence to an erroneous contour line.

[0010]

According to a first aspect of the present invention, an original image stored in an original image memory is displayed on an image display unit, a contour is designated by contour designating means, and the designated contour is converged. In the image extracting apparatus, the means for converging to a predetermined contour line by the means, and extracting a part of the original image in the converged contour line, the contour designating means is a contour of the original image on the image display unit. A pointing device for inputting a line segment that intersects with a line, an intersection point detection unit that detects an intersection point between the input line segment of the pointing device and the contour line of the original image, and the intersection point information detected by the intersection point detection unit. A polygon wheel that forms a polygonal contour line connecting the intersection information memory to be stored and the intersection point corresponding to the intersection point information in the intersection point information memory, and outputs the polygonal contour line to the contour converging means as a designated contour. And it is characterized in that formed by and a line forming means.

According to a second aspect of the present invention, the intersection point detecting means of the first aspect of the present invention detects the luminance of pixels on a line segment that is substantially orthogonal to the contour line of the original image input from the pointing device. Means, a difference calculating means for calculating a difference between the adjacent pixels of the brightness detected by the brightness detecting means, and a position of the pixel having the maximum difference calculated by the difference calculating means are detected and output as intersection point information. And a maximum value detecting means.

According to a third aspect of the present invention, the intersection point detecting means of the first aspect of the present invention is color detection for detecting the color of a pixel on a line segment that is substantially orthogonal to the contour line of the original image input from the pointing device. Means, a component difference detecting means for detecting a component difference between adjacent pixels for a specific color of the color detected by the color detecting means, and a pixel position where the component difference detected by the component difference detecting means is the maximum value And a maximum value detecting means for detecting and outputting as the intersection point information.

[0013]

In the image extracting apparatus according to the first aspect of the invention, the contour designating means operates as follows. That is, when a line segment (for example, a rough normal line) that intersects the contour line of the original image on the image display unit is input by the pointing device,
The intersection detecting means detects the intersection of the input line segment of the pointing device and the contour of the original image, the detected intersection information is stored in the intersection information memory, and the polygon contour forming means stores the intersection information in the intersection information memory. A polygonal contour line connecting the corresponding intersections is formed, and is output to the contour converging means as a designated contour.

The contour converging means converges the contour designated by the polygonal contour forming means to a predetermined contour line corresponding to the extraction target. Then, a part of the original image (that is, the extraction target image) is extracted in the contour line converged by the image synthesizing means originally possessed by the apparatus.

In the image extracting apparatus according to the invention of claim 2,
Except for the intersection detecting means, the same operation as the invention of claim 1 is performed, and the intersection detecting means operates as follows. That is, the brightness detecting means detects the brightness of the pixel on a line segment (that is, a rough normal line) substantially orthogonal to the contour line of the original image input from the pointing device, and the difference calculating means calculates the difference in brightness between adjacent pixels. Then, the maximum value detecting means detects the position of the pixel having the maximum difference, and outputs it as the intersection point information to the polygonal contour line forming means.

According to the image extracting apparatus of the third aspect,
Except for the intersection detecting means, the same operation as the invention of claim 1 is performed, and the intersection detecting means operates as follows. That is, the color detection means detects the color of a pixel on a line segment (that is, a rough normal line) that is substantially orthogonal to the contour line of the original image input from the pointing device, and the component difference detection means detects the color difference between adjacent pixels for a specific color. , The maximum value detecting means detects the position of the pixel having the maximum component difference, and outputs this as intersection point information to the polygon contour forming means.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of an image extracting device according to the present invention will be described below with reference to FIGS. In these figures, the same parts as those in FIG. 6 are designated by the same reference numerals. In FIG.
Reference numeral 30 denotes a tablet as an example of a pointing device for inputting position information to an image display unit or a computer.
The tablet 30 is coupled to a microprocessor 34 via an interface 32.

In the microprocessor 34, a ROM (read only memory) 36 storing a program for performing contour specifying processing peculiar to the present invention and a program for performing contour convergence processing similar to the conventional example. When,
An intersection information memory (for example, RAM) 38 for storing the intersection information peculiar to the present invention, an original image memory (for example, VRAM) 40 for storing the original image, and a contour memory (for example, VRAM) 40 for storing the contour information. VRAM) 42, and the original image memory 40 and the contour memory 42,
A CRT display 46, which is an example of an image display unit, is connected via the display controller 44.

The microprocessor 34 includes the RO
Based on the contour designation processing program in M36, the following processings (a) to (f) are performed. (A) A line segment display function of displaying a corresponding line segment on the CRT display 46 by inputting a line segment from the tablet 30. (B) A brightness detection function for detecting the brightness of pixels on a line segment (for example, on a general normal line that is substantially orthogonal to the contour line of the original image) input from the tablet 30.

(C) A difference calculation function for calculating the difference in brightness between adjacent pixels with respect to the brightness detected by the pixel brightness detection function of (b). (D) A maximum value detection function that compares the differences calculated by the difference calculation function of (c) above and detects the position of the pixel that gives the maximum value as an intersection. (E) A cross point information writing function of writing the cross point information detected by the cross point detection function of (d) into the cross point information memory 38.

(F) A polygon contour line forming function of connecting the intersection points corresponding to the intersection point information in the intersection point information memory 38 written in (e) to form a polygonal contour line. The brightness detection function of (b), the difference calculation function of (c), and the maximum value detection function of (d) are the line segment by the line segment display function of (a) and the contour line of the original image 12. It constitutes an intersection point detection function for detecting the intersection points of.

Next, the operation of the embodiment of FIG. 1 will be described with reference to FIGS.
Will be described together. (A) First, similarly to the conventional example, a CRT display in which image data of an original image captured by an image input device such as a television camera is stored in the original image memory 40 and is controlled by the display controller 44. An original image 12 as shown in FIG. 4A is displayed on the display screen 10 of 46.

(B) Next, by the line segment display function of the microprocessor 34 using the program in the ROM 36, based on the input from the tablet 30, (b) of FIG.
As shown in, the display screen 10 of the CRT display 46
Object to be extracted from the original image 12 (for example, a person)
14, the normal lines 50, 50, which are substantially orthogonal to the contour line 52 of
50, ... is drawn.

The outline normals 50 and 5 drawn in (b) above
0, 50, ... It is desirable to draw more as the shape of the contour line 52 is more complicated in order to improve the accuracy of the intersection information, and to prevent the background line 20 from being pulled and causing an erroneous input. Therefore, it is desirable to avoid the intersection of the strong background line 20 and the outline 52 and draw the line.

(C) Next, the brightness detection function of the microprocessor 34 using the program in the ROM 36 detects the brightness of the pixels on the approximate normal lines 50, 50, 50, .... This brightness detection is performed by, for example, the rough normal line 5 which is substantially orthogonal to the contour line 52 of the original image as shown in FIG.
For pixels 0 to 0, the brightness level represented by the length of the bar graph shown in FIG.

In FIG. 2, for convenience of illustration, the contour line 5
The size of pixels is exaggerated and displayed with respect to the thickness of 2, and the number of pixels is displayed small. In addition, in FIG. 2A, the luminance level of the shaded area on the right side of the contour line 52 is higher than the luminance level of the non-shaded area on the left side of the contour line 52.

(D) Next, the approximate normal lines 50, 50, 50 detected in (c) above are calculated by the difference calculation function of the microprocessor 34 using the program in the ROM 36.
Regarding the brightness of the upper pixel, the difference in brightness between adjacent pixels is calculated. For example, with respect to the pixels on the schematic normal line 50 shown in FIG. 2A, the difference between the adjacent pixels of the brightness level shown in FIG. 2B is calculated.

(E) Then, the maximum value detecting function and the intersection point information writing function of the microprocessor 34 using the program in the ROM 36 are used to compare the differences obtained for each rough normal line by the above (d), The maximum value is detected for each rough normal, the position of the pixel giving the maximum value is set as an intersection, the intersection information is written in the intersection information memory, and the intersection is displayed on the display screen as shown in FIG. 4B. 54, 54, 5
4, ... is displayed. For example, in FIGS. 2A and 2B, the difference in luminance between pixels is the maximum, and the position of the pixel is detected as the intersection point 54.

(F) Next, the intersection point detected in (e) is connected to the already detected intersection point (for example, the intersection point detected immediately before) by the polygon contour forming function of the microprocessor 34 using the program in the ROM 36. When the polygonal contour line 56 is formed as shown in FIG. 4C, “YES” is displayed in “Formation of polygonal contour line?”. "Add?" Is "N
O ”, and the contour command processing ends. When an intersection is added, as in the case where the formed polygonal outline 56 and the outline 52 of the extraction target 14 are largely displaced, “Are intersection added?” Becomes “YES”, and the above (b) Return to.

(G) When the intersection is added in the above (f) and the intersection is incorporated into the already formed polygonal contour line 56, is it connected to the already detected intersections adjacent to both sides of the added intersection? , Or an already detected intersection point to be connected is designated by the tablet 30, and is automatically incorporated.

(H) When the contour command processing is completed in the above (f), the microprocessor 34 executes the contour convergence processing program stored in the ROM 36, so that the polygonal contour line shown in (c) of FIG. 56 automatically converges on a predetermined contour line 58 as shown in FIG. 4D, and the contour line data is recorded in the contour memory 42.

(I) Next, the display controller 44 reads out and combines the information of the original image 12 in the original image memory 40 and the information of the contour line 58 in the contour memory 42,
The extracted image 14 as shown in FIG. 4E extracted from the original image 12 with the convergent contour line 58 is displayed.

In the above embodiment, the polygonal contour line forming means in the contour designating means detects the intersection point by the intersection point detecting means,
Although the polygonal outline is formed by connecting the detected intersection with the detected intersection each time, the present invention is not limited to this, and the intersection information necessary for forming the polygon is stored in the intersection information memory. After the process of writing into the intersection information is completed, the intersection points corresponding to the intersection point information in the intersection point information memory may be connected to form a polygonal contour line.

In the above-mentioned embodiment, the intersection detecting means is a brightness detecting means for detecting the brightness of the pixel on the rough normal line by the tablet input, and a difference calculating means for calculating the difference between the detected brightness and the adjacent pixels. The position of the pixel where the calculated difference has the maximum value is detected and is formed by the maximum value detecting means for outputting as the intersection point information, but the present invention is not limited to this, and the input line segment by the tablet and the original Any method may be used as long as it can detect an intersection where the contour of the image intersects.

For example, a color detecting means for detecting a color of a pixel on a rough normal line by tablet input, a component difference detecting means for detecting a component difference between adjacent pixels for a specific color of the detected color, and this detection. It may be formed by the maximum value detecting means for detecting the position of the pixel having the maximum component difference and outputting it as the intersection point information.

In the above-mentioned embodiment, the input line segment by the tablet is a normal line (that is, a rough normal line) substantially orthogonal to the contour line of the original image, but the present invention is not limited to this, and the contour line of the original image is not limited to this. Any line may be used as long as it intersects the line to form an intersection.

[0037]

The image extracting device according to the first aspect of the invention is
As described above, since the intersection point for forming the specified contour line is automatically determined only by inputting the line segment that intersects the contour line of the original image with the pointing device, the rough contour line of the conventional example and the Since it is not necessary to input the contour line as close as possible to the contour line of the extraction target like the outer contour line, workability can be improved.

Moreover, since the polygonal contour line connecting the intersection points detected by the intersection point detecting means is constituted as the designated contour line to the contour converging means, it is possible to perform the contour designation by the conventional rough contour line or the outer contour line. It is possible to perform contour designation with higher accuracy. Therefore, the time required for the contour convergence processing after the contour designation can be shortened, the influence of the contour designation by the manual input can be reduced, and the convergence to the wrong contour line can be prevented.

The image extracting apparatus according to the invention of claim 2 has the following effect in addition to the effect of the invention of claim 1. That is, the polygonal contour line is defined by the intersection point detection unit in the contour designation unit as the intersection point information which is the position information of the pixel on the line segment (that is, the approximate normal line) that is substantially orthogonal to the contour line of the original image. Since it is configured to be output to the forming means, it is possible to further improve the accuracy of the intersection point position information, further reduce the influence of the contour designation by manual input, and further prevent the convergence to an incorrect contour line. be able to.

The image extracting apparatus according to the invention of claim 3 has the following effect in addition to the effect of the invention of claim 1. That is, the intersection point detecting means in the contour specifying means obtains the position information of the pixel having the maximum component for a specific color of the pixel color on the line segment (that is, the approximate normal line) that is substantially orthogonal to the contour line of the original image. Since the intersection point information is output to the polygonal contour line forming means, the accuracy of the intersection point position information can be further increased, the influence of the contour designation by manual input can be further reduced, and the erroneous contour can be further reduced. The convergence to the line can be further prevented.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram showing an embodiment of an image extraction device according to the present invention.

2A and 2B are explanatory diagrams illustrating an intersection point detection function by the microprocessor of FIG. 1, in which FIG. 2A illustrates a state in which a contour line and an outline normal line thereof intersect, and FIG. 2B illustrates pixels on the outline normal line. It is an explanatory view explaining the size of the brightness level of.

3 is a flow chart for explaining the operation of the contour command means of FIG.

4A and 4B are explanatory views for explaining the operation of FIG. 1 together with a display image on a CRT display, in which FIG. 4A illustrates a state in which an original image is displayed, and FIG. 4B illustrates an outline normal line and a contour line. The state where the intersection is displayed is explained, (c) the state where the polygonal contour line is displayed, (d) the state where the contour line is converged by the contour converging means, and (e). FIG. 7 is an explanatory diagram illustrating a state where an extracted image is displayed.

FIG. 5 is a flowchart for explaining the outline of the operation of the conventional device.

6A and 6B are explanatory diagrams for explaining the operation of the conventional device together with the display image on the image display unit, in which FIG. 6A illustrates a state in which an original image is displayed, and FIG. 6B illustrates a rough contour line. The state will be described, (c) is a state in which the outline is converged by the outline converging unit, and (d) is an explanatory diagram illustrating a state in which the extracted image is displayed.

[Explanation of symbols]

10 ... Display screen, 12 ... Original image, 14 ... Extraction target (for example, person), 30 ... Tablet (an example of pointing device), 34 ... Microprocessor, 36 ...
ROM, 38 ... Intersection information memory, 40 ... Original image memory, 42 ... Contour memory, 46 ... CRT display,
50 ... Outline normal (an example of input line segment), 52 ... Original image 1
2 contour line, 54 ... intersection, 56 ... polygonal contour line, 58
… Convergent contour lines.

Claims (3)

[Claims] 1. An original image stored in an original image memory is displayed on an image display unit, a contour is designated by a contour designating means, the designated contour is converged to a predetermined contour line by the contour converging means, and the contour is converged. In the image extracting apparatus configured to extract a part of the original image within the contour line, the contour designating unit is a pointing device for inputting a line segment that intersects the contour line of the original image on the image display unit. An intersection point detecting means for detecting an intersection point between the input line segment of the pointing device and the contour line of the original image, an intersection point information memory for storing the intersection point information detected by the intersection point detecting means, and the intersection point information memory A polygonal contour line forming means for forming a polygonal contour line connecting the intersections corresponding to the intersection point information and outputting the polygonal contour line as the designated contour to the contour converging means. That image extraction apparatus. 2. The intersection detecting means includes a brightness detecting means for detecting the brightness of a pixel on a line segment substantially orthogonal to the contour line of the original image input from the pointing device, and an adjacent pixel having the brightness detected by the brightness detecting means. The difference calculation means for calculating the difference between the two and the maximum value detection means for detecting the position of the pixel having the maximum difference calculated by the difference calculation means and outputting it as intersection point information. Image extraction device. 3. The intersection detecting means includes a color detecting means for detecting a color of a pixel on a line segment substantially orthogonal to a contour line of an original image inputted from a pointing device, and a specific color of the color detected by the color detecting means. Component difference detecting means for detecting a component difference between adjacent pixels regarding color, and maximum value detecting means for detecting a position of a pixel having the maximum component difference detected by the component difference detecting means and outputting it as intersection information. Claim 1 comprising
The image extraction device described.