JPH1153561A - Device and method for searching route, generating contour curve as well as editing contour curve, and storage medium with recorded program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To exactly detect a contour through the simple operation of an opera tor. SOLUTION: The route searching device is provided with a two-point selecting means 151 for selecting two continuous points from a dot sequence data storage means 16, route search processing optimizing means 152 for calculating a parameter for optimizing a route search processing optimizing method from these two points and image data from an image data storage means 17, route searching means 153 for determining the route search processing method based on that parameter, connecting these two points, calculating route data passing the contour on the image and holding them in a route data storage means 23, and curve approximating means 154 for preparing curve data, which express the shape of the contour by the curve approximation of route data, and holding them in a curve data storage means 18. With this route searching device, and a contour curve generator to be used for this device, a contour curve editor is constituted.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a path searching apparatus and method, a contour curve generating apparatus and method, a contour curve editing apparatus and method, and those used in image processing for extracting a specific area on an image. And a recording medium storing the program.

[0002]

2. Description of the Related Art A process for extracting a specific area on an image is performed for image processing such as object recognition, tracking, and image synthesis. Various methods for extracting a region on an image have been proposed because they are fundamental and important in image processing technology, and the focus has been on automatic extraction rather than accurate extraction of a region on an image. The method can be broadly divided into a method and a method that mainly focuses on extracting a region more accurately with the input of an operator.

For example, in a process of extracting a contour of an object on an image, a conventional method proposed to extract a more accurate region by an operator's intervention input is as follows.
It is roughly classified into the following two types.

[0004] In the first method, an operator performs a deletion or addition operation on contour information automatically extracted from an image by some processing. At this time,
The operator determines that the automatically extracted contour information is unnecessary or determines the contour information that is not automatically extracted but is necessary, and performs the above operation.

[0005] Such a method is disclosed in Japanese Unexamined Patent Publication No. 3-29.
No. 1770, “Image contour detection method”,
There is a technique shown in a specific example in "Image Editing Apparatus" described in JP-A-175076.

In the second method, the operator judges a portion where the initial input information for the contour extraction processing is insufficient, and gives a more accurate input to perform the recalculation.

As such a method, Japanese Patent Laid-Open No.
2481, “Image contour detection method”
No. 155,476, entitled "Pattern Mask Creation Method", Japanese Unexamined Patent Publication No. 4-68763, "Cutout Mask Creation Method", and the document "Intelligent Scissors for
Image Composition ”(Eric N. Mortensen and Willia
mA. Barrett, Computer Graphics Proceedings, Annual
Conference Series, 1995, ACM SIGGRAPH, pp.191-19
8) is a technique shown in a specific example.

Next, each technique described in the above-mentioned documents and the like will be briefly described.

Japanese Patent Laid-Open Publication No. Hei 3-291770 discloses an "image outline detection method". In the first automatic extraction, an outline is extracted by edge detection. A technique has been disclosed in which by designating both ends, a line segment connecting them is added as a contour.

An "image editing apparatus" described in Japanese Patent Application Laid-Open No. 1-175076 discloses an image editing apparatus in which a region surrounded by a schematic line by an operator among regions in an image previously divided by color quantization is a target region. When there is an extraction error, the operator sets the corresponding area as 1
There has been disclosed a technique in which a region can be specified and modified by adding or deleting the region to or from an object region.

Also, a "method of creating a pattern mask" described in JP-A-1-155476 and JP-A-4-6876.
The “cut-out mask creation method” described in Japanese Patent Publication No. 3 is a technique similar to chroma key, and a color boundary of a contour area is determined by inputting a color on both sides of a contour, that is, one or both of a foreground color and a background color. A technique has been disclosed which enables the color input to be detected and partially changed if necessary.

Further, the above-mentioned reference "Intelligent Scisso
"rs for Image Composition" describes a technique in which a number of relay points on a contour are given to search and extract a contour point therebetween. This search method searches for a path connecting relay points so as to pass over the edge as much as possible, based on a search rule using the edge information of the image. Can be extracted. Therefore, this method has an advantage that a contour can be extracted with a relatively small number of relay points.

The "image contour detection method" described in Japanese Patent Laid-Open Publication No. Hei 3-291770 described above also searches for and extracts contour points between them by giving some relay points on the contour. This method is used when performing contour search.
The search direction is biased depending on the position of the relay point. For this reason, there is a feature that it is easy to extract a contour traveling in a desired direction by changing the relay point.

However, the conventional method described above has the following problems.

According to the first method, since the accuracy of the finally obtained contour largely depends on the accuracy of the first automatic extraction, the contour can be made more accurate by a subsequent correction operation by the operator. There is a problem that it is difficult.

For example, in the "image contour detecting method" described in Japanese Patent Application Laid-Open No. 3-291770, as described above, the contour can be manually added only by a line segment, so that the obtained contour shape lacks accuracy. There is a problem.

In the "image editing apparatus" described in Japanese Patent Application Laid-Open No. 1-175076, the area to be initially divided cannot be too small in order to prevent the correction operation from becoming complicated. Therefore, it is necessary to make the color quantization appropriately coarse. However, if the color quantization is made coarse, the shape of the region obtained by the region division and the true object outline shape are likely to be inconsistent, and the accuracy of the automatically extracted outline shape is lost.

On the other hand, the above-described second method is advantageous in terms of the accuracy of the obtained contour, because re-calculation is performed by giving a more accurate input to a region where the result of automatic extraction of the contour is not sufficient. is there. However, this method has the following problems.

The modification of the contour shape by the operator in the "method of creating a picture mask" described in JP-A-1-155476 and the "method of creating a cutout mask" described in JP-A-4-68763 is performed as described above. This is performed by changing the input of the background color or the foreground color, and recalculating the outline extraction in that part. However, the color input operation corresponds to an operation of adjusting a key level while monitoring a monitor screen in a chroma key device. Since this type of operation does not directly input information on the contour shape, it is difficult for the operator to intuitively perform correct input, and there is a problem that trial and error is required to obtain a desired contour shape.

On the other hand, "Image contour detection method" described in Japanese Patent Application Laid-Open No. 4-152481 and a document "Intelligen"
In the technique described in “t Scissors for Image Composition” (hereinafter, referred to as a conventional technique of inputting a relay point), an operator's correction operation is to operate a relay point on a contour. The contour shape is recalculated. In this method, since an input relating to a contour shape is directly provided, it is relatively easy for an operator to intuitively perform a correct input.

However, these conventional techniques also have the following problems.

In the prior art for inputting a relay point, the operator can set a desired contour by setting the relay point at or near a position where the contour is desired to pass. May not be obtained.

FIG. 28 shows an example in which it is difficult to obtain a contour desired by an operator in the conventional method of inputting a relay point.

FIG. 28A shows the shape of a contour to be extracted. In this case, the contours are a, b, c, and d.
It is assumed that the edge detection strength differs from each other. Specifically, it is assumed that a, c, and d are strongly detected by edge detection, and b is not detected so strongly. On the other hand, it is assumed that the operator grasps the entire image and determines that a, b, and c are contours to be obtained.

However, the contour to be originally obtained is not always detected strongly by edge detection.

FIG. 28B shows an example of a contour extracted by a conventional method of inputting a relay point. In any of the methods for inputting a relay point, edge detection is performed in the process, and a contour shape is determined so as to pass through the detected edge as much as possible. For this reason, as shown in FIG. 28B, when the relay point is set at the positions of p1 and p2, the edge passes through a, d, and c where edges are strongly detected, contrary to the intention of the operator. The contour e1 is output.

FIG. 28C shows that the operator
An example of an output result by a conventional method of inputting a relay point when a relay point is placed at the position of p3 in order to correct the contour extraction result of (B) is shown. Relay point p on contour b
Since 3 is set, the contour shape is corrected so as to pass in the vicinity.

However, since the result of edge detection has not changed, near the boundary between a, b, and d, FIG.
Similarly to the contour extraction result of (B), the contour is extracted so as to temporarily advance toward the edge d that is strongly detected.
At the boundary between b, c and d, the contour e2 having the wrong shape is extracted for the same reason.

After all, at the relay points p4, p5, p6, and p7, by repeating the correction as shown in FIG. 28C, as shown in FIG. The contour e3 is obtained.

As described above, in order to obtain a desired contour shape by an operator, it is often inconvenient that many relay points must be set finely.

In the prior art for inputting the above-mentioned relay point,
By setting the relay point at or near the position where the operator wants the outline to pass, the outline can be formed into a desired shape. However, in a situation described below, a desired contour may not always be obtained.

FIG. 29 shows an example of a situation where it is difficult to obtain an accurate contour with the prior art for inputting a relay point.

FIG. 29A shows a contour a to be extracted.
Is shown. Here, it is assumed that the edge of the contour a has been sufficiently detected.

FIG. 29B shows that the operators p1, p
The state of the contour shape extracted when the relay point is set at the position of 2, 2, p3, p4 is shown. Relay points p1, p2, p3
Since p4 is exactly on the contour a, the extracted contour e1 is equal to the contour a.

However, as shown in FIG. 29C, incorrect relay points p5, p6, p
The contour e2 extracted when 7, p8 is set has an inaccurate stab shape near each relay point. This is because the contour is extracted so as to pass through the detected edge as much as possible, so that the edge point is traced to the vicinity of the relay point, and the direction from the edge point closest to the relay point to the relay point rapidly changes. This is because the path to be changed is determined to be the optimum path. In order to prevent this phenomenon from occurring, the operator must accurately set the relay point on the contour as shown in FIG.

The above-mentioned "image contour detection method" searches for a route that does not necessarily pass over a relay point, but it is a process of searching for a route toward the relay point, and thus essentially has the above problem. . In this method, the above-described phenomenon is somewhat alleviated by switching the search direction to a subsequent relay point after the search has proceeded to the vicinity of the relay point so that the search point does not necessarily pass over the relay point. However, a specific method of determining the switching of the search direction to a subsequent relay point has not been specified.

[0037]

As described above,
In a conventional method for extracting a specific area on an image, in an area extraction process that focuses on extracting an area more accurately with an operator's input as an intermediary, a relay on an outline is performed. The method of inputting points has advantages over other processing methods.

However, in the method of inputting a relay point, there may be situations where it is difficult for an operator's input to be reflected in a result or an accurate relay point input is required, and it is difficult to obtain a desired contour shape. There was a problem of becoming.

The present invention has been made to solve such a problem, and a path search apparatus and method, a contour curve generation apparatus, and a path search method capable of performing accurate contour detection by a simple operation of an operator. It is an object to provide a method, a contour curve editing device and method, and a recording medium on which a program is recorded.

[0040]

According to a first aspect of the present invention, there is provided a route searching apparatus for searching for a curved path on an image, comprising: an image input unit for inputting an image; Path search processing for changing a path search parameter so as to optimize a preset path search calculation using point selection means for selecting two points on the image, and the image information and the information on the two points. It is characterized by comprising optimization means and path search means for calculating a nearby path of the contour between the two points using the optimized path search parameters.

According to a route search method of the present invention proposed to solve the above-mentioned problem, in a route search method for searching a curved route on an image, an image inputting step of inputting an image,
A point selection step of selecting two points on the image, and a path search process of changing a path search parameter so as to optimize a preset path search calculation using the image information and the information of the two points. An optimization step; and a path search step of calculating a nearby path of the contour between the two points using the optimized path search parameters.

A recording medium of the present invention proposed to solve the above-mentioned problem is a recording medium on which a path search program for searching for a curved path on an image is recorded, and the path search program inputs an image. An image input step, a point selection step of sequentially selecting two points on the image, and a preset path search calculation is optimally processed using the image information and the information of the two points, A path search processing optimizing step of changing a path search parameter each time the two points are selected; and a path search step of calculating a nearby path of a contour between the two points using the optimized path search parameters. It is characterized by having.

A contour curve generating device according to the present invention proposed to solve the above-mentioned problem is a contour curve generating device for generating a curve indicating a contour of an object on an image. Relay point sequence input means for inputting a sequence of relay points on an outline of an object on an image; and between the two adjacent points of the sequence of relay points, using the image information and the information of the two points in advance. Path search processing optimization means for changing the path search parameters so as to optimize the set path search calculation; and calculating a nearby path of the contour between each of the two points using the optimized path search parameters It is characterized by comprising a route search means and a curve generation means for generating a curve using the calculation result of the route search.

A contour curve generating method according to the present invention proposed to solve the above-mentioned problem is a contour curve generating method for generating a curve indicating the contour of an object on an image. A relay point sequence inputting step of inputting a relay point sequence on an outline of an object on an image, and, between each two adjacent points of the relay point sequence, using the image information and the information of the two points in advance. A path search processing optimization step of changing a path search parameter so as to optimize the set path search calculation, and calculating a neighboring path of a contour between each of the two points using the optimized path search parameter It is characterized by having a route searching step and a curve generating step of generating a curve using the calculation result of the path searching.

A recording medium of the present invention proposed to solve the above-mentioned problem is a recording medium on which a contour curve generating program for generating a curve indicating a contour of an object on an image is recorded. An image inputting step of inputting an image, a relaying point sequence inputting step of inputting a relaying point sequence on an outline of an object on the image, and the image information for each of two adjacent points of the relaying point sequence. A path search processing optimization step of changing a path search parameter so as to optimize a preset path search calculation using the information of the above two points, and each of the above-described steps using the optimized path search parameters. The method includes a path searching step of calculating a nearby path of a contour between two points, and a curve generating step of generating a curve using the calculation result of the path searching.

A contour curve generating apparatus according to the present invention proposed to solve the above-mentioned problem is a contour curve generating apparatus for generating a contour curve indicating a contour of an object on an image, and an image input means for inputting an image. Point sequence generating means for generating an appropriate sequence of relay points from a curve by inputting an external input curve approximating an object contour curve on the image; and inputting the generated sequence of relay points By using the image information and the information of the two points between each two points adjacent to each other,
Curve reconstruction for calculating a nearby route of the contour between each of the two points using a route search parameter optimized so as to optimize a preset route search calculation, and generating a curve using the calculation result Means.

A contour curve generating method according to the present invention proposed to solve the above-mentioned problem is a method for generating a contour curve indicating a contour of an object on an image, comprising: an image inputting step of inputting an image; A point sequence generating step of inputting an external input curve approximating an object contour curve on an image to generate an appropriate relay point sequence from the curve, and inputting the generated relay point sequence as an input and adjoining the relay point sequence About each two points,
Using the image information and the information of the two points, a route near the contour between the two points is calculated using a route search parameter optimized so as to optimize a preset route search calculation. And a curve reconstructing step of generating a curve using the calculation result.

A recording medium of the present invention proposed to solve the above-mentioned problem is a recording medium on which a program for generating a contour curve indicating a contour of an object on an image is recorded. An image input step of inputting an image, a point sequence generating step of generating an appropriate relay point sequence from a curve by inputting an external input curve approximating an object contour curve on the image, and the generated relay point sequence. As an input, between each two adjacent points of the relay point sequence,
Using the image information and the information of the two points, a route near the contour between the two points is calculated using a route search parameter optimized so as to optimize a preset route search calculation. And a curve reconstructing step of generating a curve using the calculation result.

A contour curve editing apparatus according to the present invention proposed to solve the above-mentioned problem is a contour curve editing apparatus capable of performing an additional operation on a contour curve extracted based on a sequence of relay points. Image input means for inputting, a point operation input means for operating a relay point at an arbitrary position on the image, a point sequence editing means for reflecting the point operation input to held relay point sequence data, A path search parameter that is changed so as to optimize a predetermined path search calculation using the image information and the information of the two points between two adjacent points of the held relay point sequence data. And a curve generating means for generating a curve using a result of calculation of a neighboring path of the contour between each of the two points calculated using the above.

The contour curve editing method of the present invention proposed to solve the above-mentioned problem is a contour curve editing method capable of performing an additional operation on a contour curve extracted based on a sequence of relay points. An image input step of inputting, a point operation input step of operating a relay point at an arbitrary position on the image, and a point sequence editing step of reflecting the point operation input on the held relay point sequence data, A path search that is changed so as to optimize a preset path search calculation using the image information and the information of the two points between each two adjacent points of the held relay point sequence data. A curve generating step of generating a curve using a result of calculation of a neighboring path of the contour between the two points calculated using the parameters.

The recording medium of the present invention proposed to solve the above-mentioned problem is a recording medium in which a contour curve editing program for performing an additional operation on a contour curve extracted based on a relay point sequence is recorded. The curve editing program includes an image input step of inputting an image, a point operation input step of operating a relay point at an arbitrary position on the image, and reflecting the point operation input on held relay point sequence data. In the point sequence editing step, between the two points adjacent to each other in the held relay point sequence data, the route search calculation set in advance is optimized using the image information and the information of the two points. And a curve generating step of generating a curve using a result of calculation of a neighboring path of the contour between the two points calculated using the path search parameter changed to (a).

According to the present invention described above, a path searching apparatus and method, a contour curve generating apparatus and method, a contour curve editing apparatus and method, and a program which can perform accurate contour detection by a simple operation of an operator are provided. A recorded recording medium can be provided.

[0053]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below with reference to the drawings.

First, a route searching device and a route searching method according to an embodiment of the present invention will be described.

FIG. 1 is a block diagram showing a configuration example of a main part of a route search device according to the present invention. This path finding apparatus extracts a contour of an object from an image, generates a curve indicating the contour shape, and performs a correction operation on the extracted contour to obtain a desired contour. Thus, the recalculation of the contour curve is performed.

More specifically, this route searching device has a function of searching for a curved route on an image, and has a function of selecting two points.
1. Route search processing optimization means 152, path search means 15
3. Each processing unit including the curve approximation unit 154 and the route data storage unit 23.

The two-point selecting means 151 selects and outputs two consecutive points from the point sequence data storing means 16.

The route search processing optimizing means 152 calculates parameters for optimizing the route searching method from the selected two points and the image data from the image data storage means 17 and sends them to the route searching means 153. .

The route search means 153 determines a route search processing method based on the parameters calculated by the route search processing optimization means 152, connects the two selected points, and passes a route passing on the contour in the image. Is calculated and stored in the route data storage means 23.

Here, the path data is a list of pixel coordinates representing eight near-line figures connecting the above two points. The route searching means 153 sends control information to the two-point selecting means 151 so as to select the next two points, as indicated by a broken line in FIG. Therefore, a path search is performed between points in the point sequence data storage means 16, and all of them are stored in the path data storage means 23.

When the two-point selecting means 151 selects all the points to the end of the point sequence and cannot output new two points, it sends control information to the curve approximating means 154 as shown by the broken line in FIG. The operation of the curve approximation means 154 is prompted. The curve approximating means 154 approximates the path stored in the path data storing means 23 with a curve, creates curve data representing a contour shape, and holds the curve data in the curve data storing means 18.

Next, a route search method according to the present invention will be described with reference to the configuration of the route search device shown in FIGS.

FIG. 2 is a flowchart showing a basic processing procedure of the route searching method according to the present invention.

First, in step S1, the two-point selecting means 15
1 selects and outputs two consecutive points from the point sequence data storage means 16.

Next, in step S2, optimization of the route search parameters is performed. On the basis of the two points selected by the point selection unit 151 and the image, the route search processing optimization unit 1
At 52, route search parameters are calculated.

Next, in step S3, a route search is performed. The route search means 153 determines a route search method based on the route search parameters obtained in the second step, and extracts a contour connecting the above two points in the form of an eight-neighbor route.

Then, in step S4, it is determined whether or not the route search between all the relay points has been completed. If not, the process returns to step S1, and two points shifted by one point from the two points selected last time are determined. Select and repeat steps S2 and S3. On the other hand, if it is determined in step S4 that the route search between all the relay points has been completed, the process proceeds to step S5.

In step S5, a contour curve approximating the eight neighboring routes is generated, and the process ends. Curve approximation means 1
A curve 54 approximates the path data stored in the path data storage means 23 to create curve data representing a contour shape.

Here, the curve approximating means 154 in the present embodiment can be realized by the technique disclosed in Japanese Patent Application No. 8-303687, entitled "Curve Generating Apparatus and Method". This technique is a method of obtaining a curve that approximates a coordinate list, and is one of the so-called curve approximation methods. Is also a robust and efficient method for noise.

However, the curve approximating means 154 in this embodiment may be any method as long as it is a method of approximating a series of coordinate lists with a curve, and a commonly used B-spline (sp
line) may be realized by a method such as generating a curve.

For a general curve approximation / interpolation method, see “Numerical Calculation Handbook” (Yutaka Ohno, Kazuo Isoda,
Ohmsha, 1990), "Shape processing engineering by computer display [I] [II]" (Fujio Yamaguchi, Nikkan Kogyo Shimbun, 1982), "Computer Graphics Principles and Pra
ctice "(Foly, vanDam, Feiner, and Hughes, Addison
Wesley, 1990).

Next, the path search processing optimizing means 152 and the path search means 153 will be described in more detail.

FIG. 3 is a block diagram showing a portion of the route search device according to the present invention where a route search is performed.

The feature of this route search device is that the route search process optimizing means 152 is provided so that the route search process optimized for the two input points is performed. Here, the route search processing optimization means 152
Two types of route search parameters to be described later are output.

FIG. 4 shows the internal configuration of the route search processing optimizing means 152.

The route search processing optimizing means 152 is composed of a passing cost calculation range determining means 156 and a passing cost calculation determining optimizing means 157.

The passing cost calculation range determining means 156 outputs a calculation range parameter with the selected two points as inputs.

The passing cost calculation optimizing means 157
Outputs a passing cost calculation parameter from the selected two points and the image data.

Then, these two types of route search parameters are input to the route search means 153 and used to determine a route search method.

FIG. 5 is a flowchart showing a processing procedure in the route search processing optimizing means 152.

First, in step S11, optimization of the passing cost calculation parameter is performed. The passing cost calculation optimizing means 157 calculates a passing cost calculation parameter from the selected two points (point 1 and point 2) and the image.

Next, in step S12, optimization of the passing cost calculation range parameter is performed, and the process ends. The calculation range parameter is calculated using the two points selected by the passing cost calculation range determining means 156 as inputs.

FIG. 6 is a block diagram showing, in more detail, a portion for performing processing relating to the passing cost calculation parameter in FIG.

The passing cost calculation optimizing means 157 is provided with a gradient parameter determining means 158.

The route search means 153 includes a passing cost calculating means 159, a minimum cost route calculating means 161, and a passing cost map storing means 160 for storing data called a passing cost map to be described later.

Gradient parameter determining means 158
Takes an image and two points as input, calculates a gradient parameter optimal for extracting a contour between the two points, and outputs the gradient parameter to the passage cost calculating means 159 of the path searching means 153.

The passing cost calculating means 159 calculates the image and the point 1
And the gradient parameters and the calculation range parameter output by the passing cost calculation range determining means 156 shown in FIG.

The minimum cost route calculating means 161 receives the passing cost map and the point 2 as inputs, and
Calculate a path that follows and connects the contours.

FIG. 7 is a flowchart showing a processing procedure in the route search means 153.

First, in step S21, a passing cost calculating means 159, which will be described later, calculates a passing cost map.

Next, in step S22, the minimum cost route calculating means 161 calculates the minimum cost route. Details of this operation will be described later.

FIG. 8 shows in more detail the configuration of the part that performs the processing relating to the above-mentioned passing cost calculation parameters.

In the present embodiment, the passing cost calculation parameters are two kinds of parameters necessary for calculating the gradient of an image. The first parameter is a normal vector indicating a direction perpendicular to the direction in which the contour advances, and the second parameter is a pixel value color change vector indicating a change in pixel value when observed in a direction crossing the contour.

The path searching means 153 performs processing for extracting a path on an image that passes through a place where the gradient intensity is higher as much as possible, utilizing the fact that the object contour has a higher gradient intensity.

The passing cost calculating means 159 comprises a gradient calculating means 164, a passing cost map calculating means 165, and the gradient data storing means 21 for performing the above processing.

The gradient calculating means 164 calculates the gradient of the image, and calculates the passing cost map calculating means 1.
65 calculates a passing cost map based on the obtained gradient such that the cost is reduced when passing through a place where the gradient intensity is higher. The position of point 1 is input to the passing cost map calculating means 165, and the starting point of the passing cost map is set to point 1.

Here, the gradient calculating means 164
Uses the technology disclosed by the present applicant in Japanese Patent Application No. Hei 7-134420, "Edge Detection Method and Edge Detector", to selectively detect only a gradient component of a necessary contour. The gradient calculating means 164 calculates a gradient so as to selectively extract a gradient of a contour extending so as to connect the two input points.

Next, the process of calculating the above two types of route search parameters will be described in detail.

First, the calculation of the passing cost calculation parameter and the process in which it is used will be described.

FIG. 9 is a flowchart showing a processing procedure in the gradient parameter determining means 158.

In step S31, the normal vector calculation means 162 calculates and outputs a normal vector perpendicular to the direction connecting the input points 1 and 2.

In step S32, the pixel value change vector calculation means calculates the value of 2 in the area near point 1 and point 2 on the image.
A change in the pixel value of the image when observed in a direction perpendicular to the direction connecting the points, for example, a color change direction is calculated and output as a color change vector.

Here, the passing cost map is information in which the direction in which each pixel enters the pixel position with the minimum passing cost is recorded.

FIG. 10 shows an example of a simple passing cost map.

The passing cost map is created for one range in the image specified by the passing cost calculation range parameter. Here, an example is shown in which the image is created in a certain small quadrilateral region.

In FIG. 10A, the direction in which each pixel enters the pixel position with the minimum passing cost is indicated by an arrow. This direction is calculated so that only one is determined for each pixel. The value described for each pixel is the cumulative value of the passing cost when the vehicle enters the vehicle in the minimum passing cost direction. This accumulated value is calculated from the point 1 which is the starting point of the route search to the accumulated value 0.
Is required.

FIG. 10B shows a process in which the minimum cost route calculating means 161 obtains a route when the passing cost map is obtained as shown in FIG. 10A.

The minimum cost route calculating means 161 receives the passing cost map and the point 2 as inputs, and traces the minimum passing cost direction of the passing cost map from the point 2 in reverse.
The path with the minimum cost from point 1 to point 2 is found. For example, the position shown in FIG. 10B (the accumulated cost is 22)
If there is a point 2, a path to the point 1 as shown by oblique lines is obtained by following the arrow from the point 2 (written in a direction opposite to that of FIG. 10A).

Next, a means for calculating the above-mentioned passage cost map at high speed will be described.

The calculation of the passing cost map is generally known as a kind of optimization problem, and a calculation method using dynamic programming or the like is well known. The aforementioned reference `` Intelligen
t Scissors for Image Composition "also describes a method using dynamic programming.

FIG. 11 shows an example of a description regarding a shortest path search algorithm using dynamic programming. In addition,
This description is based on the above-mentioned document "Intelligent Scissors for I
mageComposition ”, with some symbols and the like changed.

Next, the calculation of the calculation range parameter and the process of using it will be described.

FIG. 12 is a block diagram showing, in more detail, a portion for performing processing relating to the calculation range parameter in FIG.

The passing cost calculation range determining means 156 comprises distance calculating means 167, calculation range width calculating means 168, calculation range length calculating means 169, and calculation range calculating means 170.

The distance calculating means 167 receives points 1 and 2 as inputs and calculates the distance between the two points and the direction connecting the two points.

The calculation range length calculation means 169 calculates the obtained 2
The calculation range length is calculated using the distance between the points as input.

The calculation range width calculation means 168 calculates the obtained 2
The calculation range width is calculated using the distance between the points as input.

The calculation range calculation means 170 calculates calculation range parameters for determining the shape of the calculation range based on the calculation range length, the calculation range width, and the direction connecting the two points output by the distance calculation means 167. Then, it is input to the passage cost calculation means 159 of the route search means 153. The passing cost calculation means 159 sets the
Calculate the passing cost map.

FIG. 13 is a flowchart showing a processing procedure in the passing cost calculation range determining means 156 of FIG.

First, at step S41, the distance calculating means 1
67 inputs the points 1 and 2 and calculates the distance between the two points and the direction connecting the two points.

Next, in step S42, the calculation range width is calculated using the distance between the two points as an input. here,
The calculation range width calculation means 168 has therein a rule of the size of the calculation range width for the size of the distance between two points in the form of a calculation formula or a lookup table, and calculates the calculation range width according to the rule. decide. For example, even if the operator tries to set relay points in detail at short intervals in order to obtain a desired contour, the problem that the searched path meanders between the relay points and differs from the desired contour shape is a problem.
This can be solved by setting the rule for determining the calculation range width such that the smaller the distance between the two points, the smaller the width.

At this time, the calculation range width between the relay points set in detail at short intervals is determined to be very narrow, and a route search is performed so as to connect the relay points with a substantially straight line.
The contour shape desired by the operator is obtained.

Next, in step S43, the calculation range length calculation means 169 calculates the calculation range length from the distance between the two points. Here, the calculation range length is determined to be longer than the distance between two points by a predetermined margin length.

Finally, in step S44, the calculation range calculation means 170 calculates the calculation range parameters from the calculation range width, the calculation range length, and the direction connecting the two points. Here, the calculation range is a rectangle that surrounds two input points and is inclined in a direction connecting the two points, and the calculation range parameter indicates the positions of the four corners of the rectangle.

Next, the point sequence generation means 12 will be described.

The point sequence generating means 12 receives a curve as input and outputs a point sequence reproducing its shape and a reconstructed curve reconstructed by the point sequence. The generated point sequence and the reconstructed curve are used in the subsequent curve editing work by the configuration of the first embodiment.

FIG. 14 is a block diagram showing an example of the configuration of the point sequence generation means 12. As shown in FIG.

The point sequence generating means 12 includes an initial point sequence generating means 171, a curve reconstructing means 172, a difference detecting means 17
3. Point sequence editing means 14, image data storage means 17, point sequence data storage means 16, reconstruction curve data storage means 19
It is comprised including.

The initial point sequence generating means 171 generates an initial point sequence using an input curve and an image as inputs. The initial point sequence is stored in the point sequence data storage unit 16.

The curve reconstructing means 172 receives a sequence of points and an image, and generates a curve that passes through the sequence of points and passes over a contour on the image.

The difference detection means 173 receives the input curve and the reconstructed curve as inputs, and outputs the difference between the two shapes as a difference.

The point sequence editing means 14 includes a difference detecting means 173
The point sequence data in the point sequence data storage means 16 is operated on the basis of the difference output by.

FIG. 15 is a flowchart showing a processing procedure in the point sequence generation means 12.

First, in step S51, the initial point sequence generating means 171 extracts the characteristic points of the input curve, and stores the extracted characteristic points in the point sequence data storage means 16 as an initial point sequence.

The feature point extraction on the curve is described in
"Motion vector estimation method" described in JP-A-335270
Can be used.
In this method, a minute section bending angle is calculated at each point from the beginning to the end of the curve, and the sum thereof is defined as a cumulative bending angle. When the cumulative bending angle is larger than a predetermined threshold, the maximum minute section bending angle is calculated. Is a feature point. further,
Similar feature point extraction processing is performed recursively on the curve divided by the feature points.

Next, in step S52, the curve reconstructing means 172 calculates a reconstructed curve from the point sequence by the same processing as the above-described processing in the route searching apparatus according to the present invention.

Next, in step S53, the difference detecting means 1
73 calculates the difference between the input curve and the reconstructed curve. In this difference detection processing, difference detection points are taken at appropriate intervals on the reconstructed curve, and the distance from the difference detection point to the input curve is measured to obtain a difference value.

Next, at step S54, the difference detecting means 1
73 checks whether there is a difference value larger than a preset threshold value. If there is no difference value larger than the above threshold value, the processing of the point sequence generation means 12 ends, and the data in the point sequence data storage means 16 and the reconstructed curve data storage means 19 at that time are output. . on the other hand,
If there is a difference value larger than the above threshold value, step S
Go to 55.

In step S55, the point sequence editing means 14 edits the point sequence data based on the obtained difference. At each difference detection point where the difference value is determined to be larger than the threshold value, a new relay point is set at a point on the input curve closest to the difference value detection point and stored in the point sequence data storage means 16.

After the end of step S55, step S5
2, the process is repeated until the difference between the input curve and the reconstructed curve becomes smaller than the threshold value in step S54.

In the above-described processing, the initial point sequence is set to be small at first, and is appropriately added to places where the initial point sequence is insufficient. However, the initial point sequence is set to be large and unnecessary points are deleted. You may make it. In this case, the initial point sequence generating means 171 generates an initial point sequence to such an extent that the input curve can be sufficiently approximated by a broken line. Select and delete a relay point that is sufficiently smaller than.

The processing procedure shown in FIG. 15 is advantageous in that subsequent reconstruction is easy because the reconstruction is completed with fewer points. On the other hand, the processing procedure described later is a relay point for performing a path search. There is an advantage that the processing time can be shortened because the interval can be reduced.

Next, a contour curve generating device and a contour curve generating method according to an embodiment of the present invention will be described. Note that, in the following description, the same reference numerals are given to portions common to the respective portions of the above-described route search device.

FIG. 16 is a block diagram showing a configuration example of a main part of the contour curve generating device according to the present invention.

This contour curve generating device is for generating a curve indicating the contour of an object on an image, and is configured using the path search device shown in FIG.

Specifically, a point sequence correcting means 174a and a corrected point sequence data storage means 20 are added, and the input point sequence is corrected by the point sequence correcting means 174a so that a better contour curve can be obtained. And the correction point sequence data storage means 2
0 is stored. A curve generation process similar to the above-described process in the route search device according to the present invention is performed on the correction point sequence. This makes it possible to prevent the contour curve generated from the sequence of relay points in the route search from necessarily passing through the relay points.

Next, the contour curve generating method according to the present invention will be described with reference to the configuration of the contour curve generating device of FIG. 16 and FIG.

FIG. 17 is a flowchart showing a basic processing procedure in the contour curve generating method according to the present invention.

First, in step S61, the point sequence correcting means 1
74a corrects the input point sequence so that a better contour curve is obtained and stores the corrected input point sequence in the corrected point sequence data storage means 20.

After step S62, a curve is generated from the correction point sequence in the same procedure as the above-described processing procedure in the route search method according to the present invention.

That is, in step S62, the two-point selecting means 151 selects and outputs two consecutive points from the point sequence data storage means 16.

Next, in step S63, optimization of the route search parameters is performed. Based on the two points selected by the point selection unit 151 and the image, the route search processing optimization unit 152 calculates the route search parameters.

Next, in step S64, a route search is performed. The route search means 153 determines a route search method based on the route search parameters obtained in the second step,
The contour connecting the above two points is extracted in the form of an 8-neighbor path.

Then, in step S65, it is determined whether or not the route search between all the relay points has been completed. If the route search has not been completed, the process returns to step S62, and two points shifted by one point from the previously selected two points are determined. Select and repeat steps S63 and S64. On the other hand, when it is determined in step S65 that the route search between all the relay points has been completed, the process proceeds to step S67.

In step S67, a contour curve approximating the eight neighboring routes is generated, and the process ends.

Next, the contour curve generating device according to the present invention will be described in more detail.

FIG. 18 is a block diagram showing an example of the configuration of the point sequence correcting means 174a of the contour curve generating device according to the present invention.

The point sequence correcting means 174a includes a one point selecting means 175, a gradient calculating means 164, a maximum intensity point extracting means 176, and a gradient calculating range specifying means 17
7, point sequence data storage means 16, image data storage means 1
7. Gradient data storage means 21, gradient calculation range data storage means 22, and correction point sequence data storage means 20.

The one-point selecting means 175 selects the relay points from the point sequence data storage means 16 one by one in order.

Gradient calculation range designating means 177
Determines the range on the image in which the calculation performed by the gradient calculating unit 164 is performed based on the input from the operator, and stores it in the gradient calculating range data storage unit 22.

The gradient calculating means 164 receives the image and calculates the gradient of the image in the area specified by the preset gradient calculation range.

The maximum intensity point extracting means 176 extracts one point at which the gradient intensity is maximum based on the calculated gradient, and holds the extracted point in the correction point sequence data storage means 20. Further, the maximum intensity point extracting means 176 sends a control signal to the one point extracting means after the processing is completed. The one-point selecting means 175 starts selecting the next relay point according to the control signal.

FIG. 19 is a flowchart showing a processing procedure in the point sequence correcting means 174a. Here, the description will be given on the assumption that the gradient calculation range is specified in advance by the input of the operator.

First, in step S71, the one-point selecting means 1
75 selects one relay point from the point sequence data storage means 16.

Next, in step S72, the gradient calculating means 164 calculates the gradient of the area on the image specified by the preset gradient calculation range.

Next, in step S73, the maximum intensity point extracting means 176 extracts one point at which the gradient intensity becomes maximum based on the calculated gradient.

Next, in step S74, the maximum gradient intensity point is added to the correction point sequence.

Then, in step S75, it is determined whether or not the processing has been completed for all the relay points. If the processing has not been completed for all the relay points, the process returns to step S71, and the one-point selecting means 175 starts selecting the next relay point. On the other hand, when the processing for all relay points has been completed, the above processing is completed.

Next, another embodiment of the above-described contour curve generating device will be described.

FIG. 20 is a block diagram showing another configuration example (hereinafter, referred to as a second contour curve generating device) of the contour curve generating device according to the present invention.

The second contour curve generating device includes a two-point selecting means 151, a path searching processing optimizing means 152, a path searching means 153, a point sequence correcting means 174b, and a curve approximating means 15.
4 and a point sequence data storage unit 16, a correction point sequence data storage unit 20, an image data storage unit 17, a path data storage unit 23, and a curve data storage unit 18. It has the same function as.

In the second contour curve generating device, at the start of the processing, the point sequence data is copied to the correction point sequence data storage means 20 as it is. The two-point selecting unit 151 selects and outputs two consecutive points from the point sequence data storage unit 16.

The route search processing optimizing means 152 calculates parameters for optimizing the route search method from the selected two points and the image data, and sends them to the route search means 153.

The path searching means 153 determines a path searching method based on the parameters calculated by the path searching processing optimizing means 152, connects the selected two points, and calculates a path passing on the contour in the image. Then, it is stored in the route data storage unit 23.

The route searching means 153 sends control information to the two-point selecting means 151 as shown by a broken line, and selects the next two points. Therefore, the point sequence data storage means 16
The route search is performed between the points (1) and (2), and all of them are stored in the route data storage means 23.

When the two-point selecting means 151 has selected all the points up to the end of the point sequence and cannot output new two points, control information is sent to the point-sequence correcting means 174b as shown by the broken line. I have.

Point sequence correcting means 174b receiving control information
Corrects the point sequence position based on the generated route, and edits the point sequence data stored in the corrected point sequence data storage means 20. When it is determined that the correction amount is 0, the point sequence correction unit sends control information to the curve approximation unit 154 as indicated by a broken line.

The curve approximating means 154 approximates the obtained path by a curve to generate a contour curve, and stores the contour curve in the curve data storing means 18.

FIG. 21 is a flowchart showing a processing procedure in the second contour curve generating device according to the present invention.

First, in step S81, the two-point selecting means 1
Point selection by 51 is performed. The two-point selection means 151
Two consecutive points are selected from the point sequence data storage means 16 and output.

Next, in step S82, optimization of the route search parameters is performed. Based on the two points selected by the point sequence selection unit 178 and the image, the route search processing optimizing unit 152 calculates route search parameters.

Next, in step S83, a route search is performed. The route search means 153 determines a route search method based on the route search parameters obtained in the second step,
The contour connecting the above two points is extracted in the form of an 8-neighbor path.

Then, in step S84, it is determined whether or not the route search between all the relay points has been completed. If not, the flow returns to step S81 to return to the previously selected 2
Two points shifted by one point from the point are selected, and step S82,
Step S83 is repeated. On the other hand, in step S84,
If it is determined that the route search between all the relay points has been completed, the process proceeds to step S85.

In step S85, after the route search between all the relay points is repeatedly performed, the point sequence correcting means 1
74b corrects the point sequence position, and stores the corrected point sequence data storage means 2
The correction point sequence data stored in 0 is updated.

Then, in a step S86, it is determined whether or not the correction amount is equal to or larger than a predetermined threshold value. If the correction amount is equal to or larger than the threshold, steps S81 to S85 are repeated. On the other hand, if the correction amount is equal to or smaller than the threshold, the process proceeds to step 87.

In the step S87, the curve approximation means 154
Generates a contour curve by approximating the path data to a curve.

FIG. 22 is a block diagram showing a configuration example of the point sequence correcting means 174b of the second contour curve generating device.

The point sequence correcting means 174b is provided with an overlap detecting means 179, an overlapping eliminating means 180, an overlapping detection range designating means 181 and an overlapping detection range data storage means 24.

The duplication detecting means 179 receives the path data, the relay points in the correction point sequence data storage means 20, and the duplication detection range data as input, and sets a path in each of the relay points in the area indicated by the duplication detection range. Detects the pixel position where has passed twice or more, and outputs it as the overlapping range.

The duplication eliminating means 180 moves each relay point so that the above-mentioned duplication range is excluded from the route. The moved relay point position is stored in the correction point sequence data storage means 20 as a correction point.

The duplication detection range designating unit 181 determines the range on the image where the calculation performed by the duplication detection unit 179 is performed based on the input from the operator, and stores it in the duplication detection range data storage unit 24.

FIG. 23 is a flowchart showing a processing procedure in the point sequence correcting means 174b. Here, a description will be given assuming that the duplication detection range is specified in advance by an operator's input.

First, in step S91, one overlap detection range is selected.

Next, at step S92, the duplication detecting means 1
79 detects the presence or absence of overlapping of the routes. Here, if no duplication is detected, steps S93 and S94 are skipped. On the other hand, when the duplication of the route is detected in step S92, steps S93 and S94 are executed.

In step S93, the duplication elimination position is calculated.

Next, in step S94, the duplication eliminating means 1
80 moves the relay point so that the overlapping range calculated in step S93 is excluded from the route.

After step S94 is completed, step S91 is performed.
And repeat steps S92 to S94 for the next overlap detection range.

When it is determined in step S95 that detection has been completed for the duplication detection range specified in advance, the above processing is terminated.

Next, a contour curve editing apparatus and a contour curve editing method according to an embodiment of the present invention will be described. In the following description, the same reference numerals are given to the parts common to the respective parts of the above-described route search device and contour curve generating device.

FIG. 24 is a block diagram showing a configuration example of a contour curve editing device according to the present invention.

The contour curve editing device has a function of performing an additional operation such as moving or deleting the contour curve extracted based on the relay point sequence. It is configured using a generating device.

More specifically, this contour curve editing apparatus includes a curve input unit 11, a point sequence generation unit 12, and a point operation input unit 1.
3, four processing means consisting of point sequence editing means 14 and curve generating means 15 and three data storage means consisting of point sequence data storage means 16, image data storage means 17, and curve data storage means 18.

The point operation input means 13 is a pointing device such as a mouse or a tablet pen, and is a means for receiving the contour position of the object on the image designated by the operator and transmitting it to the point sequence editing means 14.

The point sequence editing means 14 changes the data stored in the point sequence data storage means 16 based on the operation input from the point operation input means 13 described above.

The curve generator 15 passes each point of the point sequence based on the point sequence data held in the point sequence data storage 16 and the image data stored in the image data storage 17. Thus, a curve that follows the contour on the image is generated, and the data is stored in the curve data storage unit 18.
The operator can deform the curve so as to pass over the desired contour by appropriately repeating necessary point operations.

Further, the curve input means 11 such as a tablet pen is a pointing device. The operator receives a locus tracing a desired contour shape and sends it to the point sequence generating means 12.

The point sequence generation means 12 generates a point sequence capable of reproducing the shape of the input trajectory based on the input trajectory and the image, and stores the data in the point sequence data storage means 16. Therefore, even when the operator inputs a desired contour with a curved shape, it is converted into a point sequence and can be edited.

The point operation input means 13 has been described as a pointing device. However, for example, by using an external storage device and a storage medium, an input point sequence previously stored in a file can be taken. It may be. Similarly, the curve input unit 11 may be configured to input an input curve from a file.

The contour curve editing apparatus may have a configuration in which the curve input means 11 and the point sequence generating means 12 are omitted.

Next, the contour curve editing method according to the present invention will be described with reference to the configuration of the contour curve editing apparatus shown in FIG. 24 and FIG.

FIG. 25 is a flowchart showing a basic processing procedure of the contour curve editing method according to the present invention.

In step S101, it is first determined whether or not the operator has instructed "end of operation". If "end of operation" has not been input, the process from step S102 is executed.

In step S102, it is determined whether the next input is a point sequence operation or a curve input. If it is a curve input, the process proceeds to step S103. If it is a point sequence operation, the process proceeds to step S104.

In step S103, the point sequence generation means 12
As a result, point sequence data is generated based on the input curve from the curve input unit 11 and is stored in the point sequence data storage unit 16.

On the other hand, in step S104, the point sequence editing means 14 reflects the point sequence operation on the point sequence data stored in the point sequence data storage means.

Then, in step S105, the curve generation means 15 generates a curve from the point sequence data.

Then, in step S105, it is determined whether or not the processing has been completed for all the point sequences. If not, the flow returns to step S101 to wait for the next input. On the other hand, if the processing for all the point sequences has been completed, the above processing is completed.

In the above description, two parameters, the pass cost calculation parameter and the calculation range parameter, are used as parameters for optimizing the route search. However, in the present invention, either one of them can function. It is. Further, in the present invention, the passing cost calculation parameter need not be a parameter for gradient detection. For example, for the calculation of the passing cost, the parameters of the Gaussian filter can be optimized based on the interval between the relay points by using the edge detection using the Laplacian filter and the Gaussian filter. The Gaussian filter allows to adjust how sensitive the edge detection is to the edge of a complicated shape. Therefore, the optimization is performed by estimating the complexity of the desired contour shape from the interval between the relay points provided by the operator.

Further, in the point sequence correcting means according to the present invention, the range in which the point sequence correction is performed can be specified by the operator. However, the specifying means may be omitted and the range may be fixed.

Furthermore, the contour curve editing apparatus according to the present invention can be configured as shown in FIG. 26 so that a point sequence is automatically put on the contour each time a curve is edited. When this configuration is used, the operator is not required to accurately set the relay point on the contour, and the work load can be reduced.

The curve generating means 15 shown in FIG.
Can use the contour curve generating device according to the present invention having the point sequence correcting means as described above.

FIG. 27 exemplifies the basic configuration of a path search device, a contour curve generating device, and a contour curve editing device according to the present invention.

Each of these devices includes a pointing device 1 and a monitor 2, which are input / output interfaces with an operator, a CPU 3, an internal storage device 4,
External storage device 5 and BUS connecting these to each other
6 is provided.

The pointing device 1 is a means for receiving an input from an operator, and a mouse, a tablet pen, or the like is used. The monitor 2 is means for displaying data such as images and graphics to the operator, and a CRT or the like is used.

The CPU 3 is a central processing unit in charge of calculation and control processing in the present invention. The internal storage device 4 is a data storage unit that temporarily stores data necessary for performing various calculation processes according to the present invention as described below.

The external storage device 5 is used as auxiliary storage means for storing data necessary for the above-mentioned calculation processing, and also as an input / output destination of data such as calculation processing results.

The route searching apparatus and method, the contour curve generating apparatus and method, and the contour curve editing apparatus and method according to the present invention described above extract a region on an image more accurately by inputting an operator's input. Therefore, it can be widely used for a process of extracting a specific area in an image.

[0228]

According to the present invention, an image and two points on the image are input, and a predetermined route search calculation method is operated optimally using the information of the two points and the image information. By optimizing the path search processing for changing the parameters of the path search, the area on the image can be more accurately extracted through the input of the operator.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration example of a main part of a route search device according to the present invention.

FIG. 2 is a flowchart showing a basic processing procedure of the route searching method according to the present invention.

FIG. 3 is a diagram for explaining a portion of the route search device according to the present invention that performs a route search.

FIG. 4 is a diagram for explaining an internal configuration of a route search processing optimization unit.

FIG. 5 is a flowchart for explaining a processing procedure in a route search processing optimization unit.

FIG. 6 is a diagram for describing a portion of a route search processing optimization unit that performs processing relating to a passing cost calculation parameter.

FIG. 7 is a flowchart for explaining a processing procedure in the route searching means.

FIG. 8 is a diagram for explaining in detail a configuration of a portion that performs processing relating to a passing cost calculation parameter using selective gradient detection.

FIG. 9 is a flowchart for explaining a processing procedure in a passing cost optimization unit.

FIG. 10 is a diagram showing an example of a passage cost map for route search.

FIG. 11 is a diagram illustrating an example of a shortest path search algorithm using dynamic programming.

FIG. 12 is a diagram for explaining in detail a portion that performs processing related to optimization of a calculation range by a path search processing optimization unit.

FIG. 13 is a flowchart illustrating a processing procedure in a calculation range optimizing unit.

FIG. 14 is a block diagram illustrating a configuration example of a point sequence generation unit.

FIG. 15 is a flowchart illustrating a processing procedure in a point sequence generation unit.

FIG. 16 is a block diagram illustrating a configuration example of a main part of a contour curve generation device according to the present invention.

FIG. 17 is a flowchart showing a basic processing procedure in the contour curve generation method according to the present invention.

FIG. 18 is a block diagram illustrating a configuration example of a point sequence correction unit of the contour curve generation device according to the present invention.

FIG. 19 is a flowchart illustrating a processing procedure in a point sequence correction unit of the contour curve generation device according to the present invention.

FIG. 20 is a block diagram illustrating a configuration example of a second contour curve generation device according to the present invention.

FIG. 21 is a flowchart showing a processing procedure in the second contour curve generating device according to the present invention.

FIG. 22 is a block diagram illustrating a configuration example of a point sequence correction unit in the second contour curve generation device according to the present invention.

FIG. 23 is a flowchart for explaining a processing procedure in a point sequence correction unit of the second contour curve generation device according to the present invention.

FIG. 24 is a block diagram illustrating a configuration example of a contour curve editing device according to the present invention.

FIG. 25 is a flowchart showing a basic processing procedure of the contour curve editing method according to the present invention.

FIG. 26 is a block diagram showing another configuration example of the contour curve editing device according to the present invention.

FIG. 27 is a diagram showing each device for configuring a path searching device, a contour curve generating device, and a contour curve editing device according to the present invention.

FIG. 28 is a diagram for describing a case where a desired contour is not correctly extracted in the related art.

FIG. 29 is a diagram illustrating a case where a desired contour is not correctly extracted in the related art.

[Explanation of symbols]

16 point sequence data storage means, 17 image data storage means, 18 curve data storage means, 23 route data storage means, 151 two-point selection means, 152 route search processing optimization means, 153 route search means, 154
Curve approximation means

 ──────────────────────────────────────────────────の Continuation of the front page (72) Inventor Hidehiro Hirase 6-7-35 Kita Shinagawa, Shinagawa-ku, Tokyo Inside Sony Corporation

Claims (34)

[Claims] An apparatus for searching for a path of a curve on an image, comprising: an image input unit for inputting an image; a point selecting unit for selecting two points on the image; Route search processing optimizing means for changing a route search parameter so as to optimize a preset route search calculation using the information; and a contour between the two points using the optimized route search parameter. And a route searching means for calculating a nearby route. 2. The apparatus according to claim 1, wherein said route search processing optimizing means includes a pass cost calculation optimizing means for minimizing a pass cost at each point on said image using positional information of two points on said image. The route search device according to claim 1, wherein 3. The passing cost calculation optimizing means includes gradient parameter determining means for determining a parameter for calculating a gradient of an image using position information of two points on the image. 3. The route search device according to claim 2, further comprising a passing cost calculation unit that calculates a passing cost of each point on the image by using the gradient calculation parameter as an input. 4. The gradient calculation parameter determination means includes: a normal vector calculation means for calculating a normal direction vector orthogonal to a straight line connecting two points on the image;
Pixel value change vector calculation means for calculating a pixel value change vector indicating a change direction of a pixel value of an image when the image moves in the normal direction in a region near a point, and the passing cost calculation means includes: Gradient calculating means for calculating a gradient component defined by the two types of vectors using a direction vector and the pixel value change vector as inputs, and a passing cost map calculating means for calculating a passing cost using the calculated gradient component The route searching device according to claim 3, comprising: 5. The path exploration processing optimizing means further includes a passing cost calculation range determining means for determining a range on an image for calculating a passing cost using position information of two points on the image. The route search device according to claim 2, wherein 6. The passing cost calculation range determining means calculates a distance and a direction of two points on the image, and a width of the passing cost calculation range extending in a direction perpendicular to the direction between the two points. 6. A route searching device according to claim 5, further comprising a calculation range width calculating means for controlling the distance according to the magnitude of the distance between the two points. 7. A path searching method for searching for a path of a curve on an image, comprising: an image inputting step of inputting an image; a point selecting step of selecting two points on the image; A path search processing optimizing step of changing a path search parameter so as to optimize a preset path search calculation using the information, and a contour between the two points using the optimized path search parameter A route searching step of calculating a nearby route of the route. 8. The path search processing optimizing step includes a pass cost calculation optimizing step of minimizing a pass cost at each point on the image calculated using position information of two points on the image. The route searching method according to claim 7, wherein: 9. The route searching method according to claim 8, wherein the route searching process optimizing step further includes a passing cost calculation range determining step of determining a range on the image where the passing cost is calculated. 10. A recording medium recording a path search program for searching for a curved path on an image, the path search program comprising: an image input step of inputting an image; and sequentially selecting two points on the image. A point selection step to be performed, and using the image information and the information of the two points, a path search parameter is changed every time the two points are selected so that a preset path search calculation is optimally processed. A recording medium comprising: a path searching process optimizing step of performing a path searching step of calculating a nearby path of a contour between the two points by using the optimized path searching parameters. 11. A contour curve generating device for generating a curve indicating a contour of an object on an image, comprising: image input means for inputting an image; and a sequence of relay points for inputting a sequence of relay points on the outline of the object on the image. Input means, and for each of two adjacent points in the relay point sequence, using the image information and the information of the two points, changing a path search parameter so as to optimize a predetermined path search calculation. Route search processing optimizing means, and each of the above two using the optimized route search parameters
A contour curve generation device comprising: a path search unit that calculates a nearby path of a contour between points; and a curve generation unit that generates a curve using the calculation result of the path search. 12. The contour curve generating device according to claim 11, further comprising a point sequence correction unit that corrects the relay point sequence based on information near the relay point sequence. 13. The point sequence correcting means compares the gradient in the vicinity of each relay point with the gradient calculating means for calculating the gradient of the image near each relay point in the input relay point sequence, and obtains the maximum gradient intensity. 13. The contour curve generating apparatus according to claim 12, further comprising a maximum intensity point extracting means for detecting the point (a), wherein the path searching means performs a path search so as to pass through a position where the gradient intensity on the image is large. 14. The point sequence correction means, comprising: an input relay point sequence and a route output from the route search means as inputs, and a duplication detection means for detecting a route duplication occurring near the input relay point sequence; 13. The contour curve generation device according to claim 12, further comprising: an overlap eliminating unit that corrects a position of the input relay point sequence so as to eliminate the detected overlapping route. 15. The contour curve generating apparatus according to claim 12, wherein said point sequence correcting means includes an overlapping detection range specifying means for specifying a correction processing range in each relay point sequence. 16. A contour curve generating method for generating a curve indicating a contour of an object on an image, comprising: an image inputting step of inputting an image; and a relay point sequence for inputting a sequence of relay points on the outline of the object on the image. An input step, and for each of two adjacent points in the relay point sequence, using the image information and the information of the two points, changing a path search parameter so as to optimize a predetermined path search calculation. Using the optimized path search parameters, and using the optimized path search parameters
A contour curve generating method, comprising: a path searching step of calculating a nearby path of a contour between points; and a curve generating step of generating a curve using the calculation result of the path searching. 17. The contour curve generating method according to claim 16, further comprising a point sequence correction step of correcting the relay point sequence based on information near the relay point sequence. 18. A recording medium recording a contour curve generating program for generating a curve indicating a contour of an object on an image, wherein the contour curve generating program comprises: an image inputting step of inputting an image; A relay point sequence inputting step of inputting a relay point sequence on an outline of an object; and a preset between the two adjacent points of the relay point sequence using the image information and the information of the two points. A path search processing optimizing step of changing a path search parameter so as to optimize the path search calculation; and each of the above two steps using the optimized path search parameter.
A recording medium comprising: a path searching step of calculating a nearby path of a contour between points; and a curve generating step of generating a curve using the calculation result of the path searching. 19. A contour curve generating apparatus for generating a contour curve indicating a contour of an object on an image, comprising: an image input means for inputting an image; and an external input curve approximating the object contour curve on the image. A sequence of point sequence generating means for generating an appropriate sequence of relay points from a curve, and the image information and the two points between each two adjacent points of the sequence of relay points, with the generated sequence of relay points as inputs. Using the information and the route search parameters optimized so as to optimize the preset route search calculation, a nearby route of the contour between each of the above two points is calculated, and a curve is formed using the calculation result. And a curve reconstructing means for generating the contour curve. 20. The point sequence generating means has a function of sampling a relay point on a curve with sufficient fineness, and the point sequence editing means operates while the detected difference keeps a certain value or less. 20. The contour curve generating apparatus according to claim 19, wherein the editing is performed so as to reduce the number of relay points in the point sequence. 21. The point sequence generating means has a function of sampling characteristic points on a contour curve, and the point sequence editing means determines whether or not the detected difference has a predetermined value or more. 3. Editing is performed to increase the number of relay points.
0. The contour curve generation device according to 0. 22. A method for generating a contour curve indicating a contour of an object on an image, comprising: an image input step of inputting an image; and an external input curve approximating the object contour curve on the image as an input. A point sequence generating step of generating an appropriate relay point sequence; and, using the generated relay point sequence as an input, the image information and the information of the two points between each two adjacent points of the relay point sequence. Using the route search parameters optimized so as to optimize the preset route search calculation, a nearby route of the contour between the two points is calculated, and a curve is generated using the calculation result. A contour reconstructing step. 23. In the point sequence generating step, the relay points on the curve are sampled with sufficient fineness, and the point sequence editing means performs the processing in the point sequence while the detected difference is kept below a certain predetermined value. 23. The contour curve generating method according to claim 22, wherein editing is performed so as to reduce the number of relay points. 24. In the point sequence generating step, the characteristic points on the contour curve are sampled, and in the point sequence editing step, the number of relay points in the point sequence is increased while the detected difference keeps a predetermined value or more. 24. The contour curve generating method according to claim 23, wherein the editing is performed as follows. 25. A recording medium recording a program for generating a contour curve indicating the contour of an object on an image, wherein the program for generating the contour curve comprises: an image inputting step of inputting an image; A point sequence generating step of generating an appropriate sequence of relay points from a curve by inputting an external input curve approximating an object contour curve; and each of two adjacent points of the sequence of relay points using the generated sequence of relay points as input. For the interval, using the above-described image information and the information of the two points, using a path search parameter optimized so as to optimize a preset path search calculation, a route near the contour between the two points is used. And a curve reconstructing step of generating a curve using the calculation result. 26. A contour curve editing device capable of performing an additional operation on a contour curve extracted based on a sequence of relay points, comprising: image input means for inputting an image; Point operation input means for operating a position; point sequence editing means for reflecting the point operation input to the held relay point sequence data; and between each two adjacent points of the held relay point sequence data. Calculating a neighboring route of a contour between each of the two points calculated using a path search parameter changed so as to optimize a predetermined path search calculation using the image information and the information of the two points. And a curve generating means for generating a curve using the result. 27. The curve generating means includes a curve generating means having a point sequence correcting means for correcting the generated contour curve by correcting an input relay point sequence, and wherein the holding is performed before and after the curve is generated. 27. The contour curve editing device according to claim 26, wherein the performed relay point sequence is not changed. 28. The curve generating means includes a curve generating means having a point sequence correcting means for correcting the generated contour curve by correcting an input relay point sequence, and the point sequence correcting after the curve generation. 27. The contour curve editing apparatus according to claim 26, wherein the point sequence corrected by the means is newly stored. 29. Curve input means for inputting a curve shape existing at an arbitrary position on an image, point sequence generating means for generating an appropriate relay point sequence from the input curve, and generating the generated relay point sequence As an input, the image information and the two points on the image between each two adjacent points are input.
Using the information on the points, the path search processing is optimized to change the path search parameters so as to optimize the path search calculation set in advance, and between the two points using the optimized path search parameters Calculate a path near the contour, generate a curve using the calculation result of the path search, detect a difference between the generated reconstructed curve and the input curve, and calculate the relay point according to the detected difference. 27. The contour curve editing device according to claim 26, further comprising: a point sequence generation unit configured to generate a relay point sequence by performing an editing operation on the sequence. 30. A contour curve editing method capable of performing an additional operation on a contour curve extracted based on a sequence of relay points, comprising: an image inputting step of inputting an image; A point operation inputting step of operating a position; a point sequence editing step of reflecting the point operation input to the held relay point sequence data; and a point sequence editing process between the two adjacent points of the held relay point sequence data. , A neighboring route of the contour between each of the two points calculated using a route search parameter changed so as to optimize a preset route search calculation using the image information and the information of the two points A curve generation step of generating a curve using the calculation result. 31. The curve generation step includes a curve generation step including a point sequence correction step of correcting the generated contour curve by correcting an input relay point sequence, and before and after the curve generation. 31. The contour curve editing method according to claim 30, wherein the held relay point sequence is not changed. 32. The curve generation means includes a curve generation step including a point sequence correction step of correcting the generated contour curve by correcting the input relay point sequence, and the correction is performed after the curve generation. 31. The contour curve editing method according to claim 30, wherein a point sequence is newly held. 33. A curve inputting step of inputting a curve shape existing at an arbitrary position on an image; a point sequence generating step of generating an appropriate relay point sequence from the input curve; As an input, the image information and the two points on the image between each two adjacent points are input.
Using the information on the points, the path search processing is optimized to change the path search parameters so as to optimize the path search calculation set in advance, and between the two points using the optimized path search parameters Calculate a path near the contour, generate a curve using the calculation result of the path search, detect a difference between the generated reconstructed curve and the input curve, and calculate the relay point according to the detected difference. 31. The contour curve editing method according to claim 30, further comprising a step of generating a relay point sequence by adding an editing operation to the sequence. 34. A recording medium recording a contour curve editing program for performing an addition operation to a contour curve extracted based on a relay point sequence, wherein the contour curve editing program comprises: an image inputting step of inputting an image; A point operation input step of operating a relay point at an arbitrary position on the image; a point sequence editing step of reflecting the point operation input to held relay point sequence data; and the held relay point sequence Between each two points adjacent to each other in the data, calculated using a path search parameter changed so as to optimize a path search calculation set in advance using the image information and the information of the two points. A curve generating step of generating a curve using a result of calculation of a neighboring path of a contour between each two points.