WO2012114386A1

WO2012114386A1 - Image vectorization device, image vectorization method, and image vectorization program

Info

Publication number: WO2012114386A1
Application number: PCT/JP2011/001106
Authority: WO
Inventors: 進也田口
Original assignee: 三菱電機株式会社
Priority date: 2011-02-25
Filing date: 2011-02-25
Publication date: 2012-08-30

Abstract

An image vectorization device is provided with: an edge detection unit (1) for detecting edges present in a raster image and identifying the degrees of importance of the edges in the raster image; and a region dividing unit (2) for dividing the raster image into a plurality of partial regions by using the edges with the highest degree of importance identified by the edge detection unit (1) among the edges detected by the edge detection unit (1), repeating a process for dividing the partial regions into a plurality of another partial regions by sequentially using the edges with a lower degree of importance than the edges with the highest degree of importance, and thereby hierarchizing the partial regions after being divided. A color approximation unit (3) approximates, per each of the partial regions hierarchized by the region dividing unit (2), the pixel values indicating the colors of pixels constituting the partial regions by using continuous functions.

Description

Image vectorization apparatus, image vectorization method, and image vectorization program

The present invention relates to, for example, an image vectorization apparatus, an image vectorization method, and an image vectorization program for converting a raster image into a vector image.

For example, the following Non-Patent Document 1 proposes an image vectorization method for the purpose of facilitating image editing and preventing blurring and collapse of the shape even when the image is enlarged or reduced.
That is, the following Non-Patent Document 1 describes an image that realizes image vectorization by dividing a raster image into mesh-like partial areas and approximating pixel values in each partial area with a bicubic surface. A vectorization device is disclosed.

Here, “raster image” refers to an image represented by a set of colored points, and “vector image” represents an equation in which the image is represented by dots, lines, and planes, and color information is a parametric equation. This refers to the image represented by.
“Image vectorization” refers to converting a raster image into a vector image.
“Point” refers to a two-dimensional coordinate value, a three-dimensional coordinate value, or an N-dimensional coordinate value, “Line” refers to a straight line or curve connecting two points, and “Surface” is surrounded by a plurality of lines. It refers to the area that is.

Since the conventional image vectorization apparatus is configured as described above, image vectorization is performed so that a raster image can be faithfully reproduced. For this reason, when the number of partial areas is large, the number of parameters expressing the position information of each partial area is increased, resulting in an increase in the size of vectorized image data.
Also, in texture mapping using a graphic programmable shader, when rendering an image, random access to the vectorized image data (without reconfiguring the entire image, the pixel value of the specified texture coordinate is changed. However, since the pixel values in each partial area are approximated by a bicubic surface, there is a problem that it is difficult to calculate random access.

The present invention has been made to solve the above-described problems, and can dynamically change the amount of data used for image data in accordance with the reduction in scale or viewpoint, and can be applied to image data at random. An object is to obtain an image vectorization apparatus, an image vectorization method and an image vectorization program which can be accessed.

An image vectorization apparatus according to the present invention detects an edge existing in a raster image, determines edge importance in the raster image, and detects an edge detected by the edge detection unit. Among them, the edge having the highest importance determined by the edge detection means is used to divide the raster image into a plurality of partial areas, and the edges having lower importance than the edges are used in order, By repeating the process of dividing the partial area into a plurality of partial areas, there is provided an area dividing means for hierarchizing the divided partial areas, and the color approximating means for each partial area hierarchized by the area dividing means The pixel value indicating the color of the pixels constituting the partial area is approximated by a continuous function.

According to the present invention, the edge existing in the raster image is detected, the edge detection means for determining the importance of the edge in the raster image, and the edge detected among the edges detected by the edge detection means Using the edge with the highest importance determined by the detection means, the raster image is divided into a plurality of partial areas, and edges having a lower importance than the edges are used in order, so that the plurality of partial areas are divided. By repeating the process of dividing into partial areas, an area dividing means for hierarchizing the divided partial areas is provided, and the color approximating means configures the partial areas for each partial area hierarchized by the area dividing means. Since the pixel value indicating the color of the current pixel is approximated by a continuous function, the amount of image data used can be changed dynamically according to the scale, viewpoint switching, etc. It is possible to, the effect capable of random access to image data.

It is a block diagram which shows the image vectorization apparatus by Embodiment 1 of this invention. It is a flowchart which shows the processing content (image vectorization method) of the image vectorization apparatus by Embodiment 1 of this invention. It is explanatory drawing which shows the edge detection process by the edge detection part 1, and importance determination process. It is explanatory drawing which shows the area division process by the area division part 2, and the hierarchization process of a partial area. It is explanatory drawing which shows the area division method of the area division part. It is explanatory drawing which shows another area dividing method of the area dividing part. It is explanatory drawing which shows the attribute setting process of the edge etc. which surround the partial area | region by the color approximation part 3, and the sampling process of a pixel value. It is explanatory drawing which shows the local coordinate system (U, V) of the surface with a sampling point. It is explanatory drawing which shows the structural example of vector image data. It is explanatory drawing which shows the structural example of an area | region access table. It is explanatory drawing which shows the drawing process of the image by a car navigation apparatus.

Hereinafter, in order to explain the present invention in more detail, modes for carrying out the present invention will be described with reference to the accompanying drawings.
Embodiment 1 FIG.
FIG. 1 is a block diagram showing an image vectorization apparatus according to Embodiment 1 of the present invention.
In FIG. 1, an edge detection unit 1 is composed of, for example, a semiconductor integrated circuit in which a CPU, MPU, or GPU (Graphics Processing Unit) is mounted, or a one-chip microcomputer, and an edge existing in a raster image. Is detected, the importance of the edge in the raster image is determined, and edge information indicating the edge and importance information indicating the importance of the edge are output to the region dividing unit 2. The edge detector 1 constitutes an edge detector.

The area dividing unit 2 is composed of, for example, a semiconductor integrated circuit on which a CPU, MPU, or GPU is mounted, or a one-chip microcomputer. Among the edges detected by the edge detecting unit 1, the edge dividing unit 2 Use the edge with the highest importance indicated by the output importance information to divide the raster image into multiple partial areas, and use the edges that are less important than the edges in order to By repeating the process of dividing into a plurality of partial areas, a process of hierarchizing the divided partial areas is performed. The area dividing unit 2 constitutes an area dividing means.

The color approximating unit 3 is composed of, for example, a semiconductor integrated circuit on which a CPU, MPU, or GPU is mounted, or a one-chip microcomputer. For each partial region hierarchized by the region dividing unit 2, the partial region is A process of approximating the pixel value indicating the color of the constituent pixels with a continuous function is performed. The color approximating unit 3 constitutes color approximating means.

The vector image data storage unit 4 is composed of a storage device such as a RAM or a hard disk, for example. The vector image data generated by the image vectorization device (for example, region information indicating partial areas hierarchized by the region dividing unit 2) Or an area access table indicating the correspondence between the image coordinates of the raster image and the partial areas hierarchized by the area dividing unit 2.

In the example of FIG. 1, it is assumed that each of the edge detection unit 1, the region division unit 2, and the color approximation unit 3 that is a component of the image vectorization device is configured by dedicated hardware. When the vectorization apparatus is configured by a computer, an image vectorization program describing the processing contents of the edge detection unit 1, the region division unit 2, and the color approximation unit 3 is stored in a computer memory, and the CPU of the computer The image vectorization program stored in the memory may be executed.
FIG. 2 is a flowchart showing the processing contents (image vectorization method) of the image vectorization apparatus according to Embodiment 1 of the present invention.

Next, the operation will be described.
When the raster image is input, the edge detection unit 1 detects an edge existing in the raster image (step ST1).
When the edge detection unit 1 detects an edge present in the raster image, the edge detection unit 1 determines the importance of the edge in the raster image (step ST2).
Here, FIG. 3 is an explanatory diagram showing edge detection processing and importance determination processing by the edge detection unit 1.
Hereinafter, the edge detection processing and importance determination processing by the edge detection unit 1 will be described in detail.

When a raster image is input, the edge detection unit 1 reduces the raster image to a plurality of scales as shown in FIG. 3A before detecting an edge existing in the raster image. A plurality of reduced images are generated.
In the example of FIG. 3A, a reduced image having a size that is ½ of a raster image and a reduced image having a size that is ¼ are generated.
When the raster image is reduced to a plurality of scales to generate a reduced image, the edge detection unit 1 detects edges existing in the plurality of scale images (including the original raster image). That is, an edge existing in a full-size raster image, an edge existing in a reduced image of 1/2 size, and an edge existing in a reduced image of 1/4 size are detected.
Note that the edge detection can use an existing method such as the Canny method, for example.

When the edge detection unit 1 detects an edge existing in a plurality of reduced images, the edge detection unit 1 determines the importance of the edge.
For example, an edge that roughly captures the outline characteristics of an object appearing in the image is determined to be a highly important edge, and an edge that captures details of the object outline in the image is a low importance edge. Is determined.

Specifically, as shown in FIG. 3B, the edge present in the reduced image of 1/4 size which is the smallest scale image is determined to be the edge having the highest importance. Then, importance 3 (the highest importance) is assigned to the edge.
Next, among the edges existing in the reduced image of ½ size, the edges that are not present in the reduced image of ¼ size (edges not assigned importance 3) are The edge is determined to be an intermediate edge, and importance 2 (medium importance) is assigned to the edge.
Finally, of the edges existing in the original raster image, the edges that are not present in the reduced images of 1/2 and 1/4 size (edges that are not given importance 3 or 2) ) Is determined to be the edge with the lowest importance, and importance 1 (the lowest importance) is assigned to the edge.

Thereby, the highest importance level 3 is given to the edges detected from all the scale images, and the lowest importance level 1 is given to the edges detected only from the original size raster image.
When the edge detection unit 1 performs the edge detection process and the importance determination process, the edge detection unit 1 obtains edge information indicating an edge existing in each scale image and importance information indicating the importance assigned to the edge. The data is output to the area dividing unit 2.

Upon receiving edge information and importance information from the edge detection unit 1, the area dividing unit 2 refers to the edge information and importance information, divides the raster image into a plurality of partial areas (step ST3), The process of dividing the partial area into a plurality of partial areas is repeatedly performed (steps ST4 and ST5).
That is, the region dividing unit 2 refers to the importance level information output from the edge detection unit 1 and identifies an edge of importance level 3 among the edges detected by the edge detection unit 1. The raster image is divided into a plurality of partial regions using the edges (step ST3).
Next, with reference to the importance level information, an importance level 2 edge that is one level lower than the importance level 3 is identified, and the above-mentioned partial image (the importance level 3 edge) is used by using the importance level 2 edge. Is divided into a plurality of partial areas (steps ST4 and ST5).
Next, with reference to the importance information, an edge of importance 1 that is one less important than importance 2 is specified, and the partial image (importance 3, 2) is used using the edge of importance 1. Are divided into a plurality of partial areas (steps ST4 and ST5).
Since there is no edge having a lower importance level than the importance level 1, the repetition process of steps ST4 and ST5 is terminated.

Hereinafter, the processing content of the area dividing unit 2 will be specifically described.
FIG. 4 is an explanatory diagram showing the area dividing process and the partial area hierarchizing process by the area dividing unit 2.
First, the region dividing unit 2 uses a “edge of importance 3” having the highest importance (see FIG. 4A) to perform a raster image dividing process, so that a plurality of raster images are obtained. Are divided into partial regions R1, R2, R3, and R4 (see FIG. 4B).

Here, the raster image dividing process will be described in detail.
FIG. 5 is an explanatory diagram showing a region dividing method of the region dividing unit 2.
As shown in FIG. 5, an example of region division when two edges (1) and (2) exist in a raster image will be described.
First, as shown in FIG. 5A, the area dividing unit 2 obtains inflection points, branch points (including intersection points), and end points of edges existing in the image.
In the example of FIG. 5, the end point of the edge (1), the intersection of the edge (1) and the edge (2), and the inflection point of the edge (2) are obtained.

Next, as shown in FIG. 5B, the area dividing unit 2 obtains area dividing lines passing through horizontal or vertical edges existing in the image, and divides the image by the area dividing lines.
In the example of FIG. 5, since the edge (1) which is a horizontal edge exists in the image, the area dividing line passing through the edge (1) is obtained. In the example of FIG. 5, no vertical edge exists in the image.

Next, as shown in FIG. 5C, the area dividing unit 2 obtains a horizontal or vertical area dividing line passing through the end points of the edges, and divides the image by the area dividing line.
In the example of FIG. 5, since the end point of the edge (1) exists in the image, a vertical area dividing line passing through the end point of the edge (1) is obtained. The horizontal area dividing line passing through the end point of the edge (1) overlaps with the area dividing line in FIG.

Next, as shown in FIG. 5D, the area dividing unit 2 obtains a horizontal or vertical area dividing line passing through the intersection of the edges, and divides the image by the area dividing line.
In the example of FIG. 5, since the intersection of the edge (1) and the edge (2) exists in the image, a vertical area dividing line passing through the intersection of the edge (1) and the edge (2) is obtained. The horizontal area dividing line passing through the intersection of the edge (1) and the edge (2) overlaps with the area dividing line in FIG.

Next, as shown in FIG. 5E, the region dividing unit 2 obtains a horizontal or vertical region dividing line passing through the inflection point of the edge, and divides the image by the region dividing line.
In the example of FIG. 5, since the inflection point of the edge (2) exists in the image, the horizontal area dividing line passing through the inflection point of the edge (2) and the vertical area dividing line are obtained. Yes.
The area dividing unit 2 repeatedly performs the operations shown in FIGS. 5B to 5E until the number of curved edges becomes one in each rectangular partial area that is an image after division.
In the operations shown in FIGS. 5B to 5E, when a new area dividing line is newly added, the existing area dividing line may be crossed or may not be crossed.

Finally, as shown in FIG. 5F, the region dividing unit 2 obtains an approximate curve by approximating a diagonal line or a curve included in the partial region to a quadratic curve or a cubic curve. The internal line information indicating the type of function indicating the approximate curve and the parameter of the function is set as one of the area information (attribute information) of the partial area.
In the example of FIG. 5F, the types of diagonal lines and curves included in the rectangular partial region are not limited, but the region division may be performed by limiting the types of curves to a certain pattern.

FIG. 6 is an explanatory diagram showing another area dividing method of the area dividing unit 2.
As shown in FIG. 5, the image may be divided into a plurality of partial areas using inflection points, branch points (including intersections), and end points of edges existing in the image. The image may be divided into a plurality of partial regions using any one of the region pattern (1), the region pattern (2), and the region pattern (3) as shown in A).

Here, the area pattern (1) is an area not including a dividing line, and the area pattern (2) is an area including a dividing line described by a curve or a straight line passing through the upper right corner and the lower left corner.
The area pattern (3) is an area including a dividing line described by a curve or a straight line passing through the upper left corner and the lower right corner.
For example, as shown in FIG. 6B, the region dividing unit 2 matches two edges (1) and edge (2) when two edges (1) and edge (2) exist in the image. Then, for each small region that is a partial region, one of the region pattern (1), the region pattern (2), and the region pattern (3) is selected (for example, a small region that does not have edges (1) and (2)). In the area, the area pattern (1) is selected, and in the small area where the edge (2) exists, the area pattern (2) or the area pattern (3) is selected according to the direction of the curve or the straight line). By combining the region patterns of the selected small regions, a region division result as shown in FIG. 6C is obtained.
By expressing each partial region with any one of the three region patterns, the types of curves in each partial region can be limited, so that the advantage of reducing the number of parameters of the curve can be obtained.

The area dividing unit 2 uses the “importance 3 edge” having the highest importance (see FIG. 4A) to divide the raster image into partial areas R1, R2, R3, and R4. Using the “importance 2 edge” having a high importance (see FIG. 4A), the partial area is divided into a plurality of partial areas by performing the division process (see FIG. 4). (See (B)).
In the example of FIG. 4, since an edge of importance 2 exists in the partial region R4, the partial region R4 is divided into partial regions R5, R6, and R7.

The region dividing unit 2 uses the “edge of importance 2” (see FIG. 4A) to divide the partial region R4 into the partial regions R5, R6, and R7, and the “importance” having the lowest importance is obtained. By using the edge of “1” (see FIG. 4A), the partial area is divided to be divided into a plurality of partial areas (see FIG. 4B).
In the example of FIG. 4, since the edge of importance 1 crosses the partial regions R1, R2, and R3, the partial region R1 is divided into the partial region R8 and the partial region R11, and the partial region R2 is divided into the partial region R9 and the partial region. Divide into R12. Further, the partial region R3 is divided into a partial region R10 and a partial region R13.

When the area dividing process is completed, the area dividing unit 2 performs a hierarchizing process on the divided partial areas R1 to R13 (step ST6).
In the example of FIG. 4, the partial areas R1, R2, R3, and R4 divided using only the “importance 3 edge” having the highest importance belong to the hierarchy (1) (the highest hierarchy). And the hierarchy number indicating the hierarchy (1) is assigned to the partial regions R1, R2, R3, and R4.
The subregions R5, R6, and R7 divided by adding the “edge of importance 2” are defined as belonging to the hierarchy (2) (middle hierarchy), and the hierarchy number indicating the hierarchy (2) Is applied to the partial regions R5, R6, and R7.
The partial areas R8, R9, R10, R11, R12, and R13 divided by adding the “edge of importance 1” are defined as belonging to the hierarchy (3) (the lowest hierarchy), and the hierarchy ( 3) is assigned to the partial areas R8, R9, R10, R11, R12, and R13.

When the region dividing unit 2 assigns the layer numbers to the partial regions R1 to R13, the partial regions R1 to R13 belonging to the layers (1) to (3) are linked to each other between the corresponding partial regions. Do.
For example, since the partial regions R5, R6, and R7 are regions divided from the partial region R4, there is a correspondence between the partial regions R5, R6, and R7 and the partial region R4, as shown in FIG. Thus, the partial region R4 and the partial regions R5, R6, and R7 are linked.

Further, since the partial areas R8, R11 are areas divided from the partial area R1, there is a correspondence between the partial areas R8, R11 and the partial area R1, and as shown in FIG. The region R1 and the partial regions R8 and R11 are linked.
Similarly, since the partial regions R9 and R12 are regions divided from the partial region R2, there is a correspondence between the partial regions R9 and R12 and the partial region R2, and as shown in FIG. The partial area R2 and the partial areas R9 and R12 are linked.
Similarly, since the partial regions R10 and R13 are regions divided from the partial region R3, there is a correspondence between the partial regions R10 and R13 and the partial region R3, and as shown in FIG. The partial area R3 and the partial areas R10 and R13 are linked.

When the partial area hierarchization processing by the area dividing unit 2 is finished, the color approximating unit 3 displays, for each partial area hierarchized by the area dividing unit 2, a pixel value indicating the color of the pixel constituting the partial area Is approximated by a continuous function.
Here, FIG. 7 is an explanatory diagram showing attribute setting processing such as a side surrounding a partial region and sampling processing of pixel values by the color approximation unit 3.
FIG. 8 is an explanatory diagram showing a local coordinate system (U, V) of sampling points.
Hereinafter, the processing content of the color approximation unit 3 will be described in detail with reference to FIGS. 7 and 8.

For each partial region hierarchized by the region dividing unit 2, the color approximating unit 3 sets the attribute of the side surrounding the partial region to “continuous side” or “discontinuous side” (step ST7).
That is, the color approximating unit 3 determines whether or not an edge exists on the side surrounding the partial region, and sets the attribute of the side on which the edge exists to “discontinuous side”. On the other hand, the attribute of the side where no edge exists is set to “continuous side”.
For example, when attention is paid to the partial region R1 shown in FIG. 4, since there is an edge at the boundary between the partial region R1 and the partial region R2 (see FIGS. 4A and 4B), there is an edge on the right side of the partial region R1. Exists and the attribute of the side is set to “discontinuous side”.
On the other hand, since there is no edge at the boundary between the partial region R1 and the partial region R4 (see FIGS. 4A and 4B), no edge exists on the lower side of the partial region R1, and the attribute of the side is Set to “continuous edge”.

Here, the side attribute surrounding the partial area is set by the presence or absence of the edge, but the side attribute may be set based on the continuity of the color near the side surrounding the partial area.
That is, the color approximating unit 3 determines the difference between the pixel value inside the partial area (the pixel value indicating the color of the pixel constituting the partial area) and the pixel value in the peripheral area (adjacent area) of the partial area. If the difference between the pixel values is equal to or greater than a predetermined value, the attribute of the side surrounding the partial area is set to “discontinuous side”.
On the other hand, if the difference between the pixel values is less than the predetermined value, the attribute of the side surrounding the partial area is set to “continuous side”.
As a method for determining the continuity of pixel values, for example, the Euclidean distance of a set of pixel values around adjacent sides may be calculated.

For example, when attention is paid to the partial area Z1 shown in FIG. 7A (inside the red area), for example, the upper area of the partial area Z1 is yellow, the right area is blue, and the red of the partial area Z1. And different. For this reason, the attribute of the upper side and the right side of the partial region Z1 is set to “discontinuous side”.
On the other hand, since the lower and left regions of the partial region Z1 are the same red, the attribute of the lower and left sides of the partial region Z1 is set to “continuous side”.
The color approximating unit 3 always sets the attribute of the diagonal line or curve included in each partial region to “discontinuous side”.

When the color approximating unit 3 sets the attribute of the oblique line or curve included in the partial region and the sides surrounding the partial region, the color approximating unit 3 determines the pixel value to be sampled according to the setting state of the attribute (step ST8). .
For example, focusing on the partial area Z1 shown in FIG. 7A, the attributes of the upper and right sides of the partial area Z1 are “discontinuous sides”, and the attributes of the lower and left sides are “continuous sides”.
The partial area Z1 includes a diagonal line L, and the attribute of the diagonal line L is “discontinuous side”.
For this reason, the partial region Z1 is divided into an upper part and a lower part by an oblique line L.

Since the upper part of the partial area Z1 is surrounded by a line having the attribute of “discontinuous side”, when approximating the pixel value indicating the color of the pixels constituting the upper part with a continuous function, the lower part of the partial area Z1 In addition, it is desirable not to mix with the colors in the upper and right areas of the partial area Z1. For this reason, only the pixel value of the pixel constituting the upper part is determined as the pixel value to be sampled. Specifically, the pixel value at the position indicated by “●” in FIG. 7B is determined as a sampling target.

Since the attribute of the lower side and the left side of the lower part of the partial area Z1 is “continuous side”, when the pixel value indicating the color of the pixels constituting the lower part is approximated by a continuous function, the lower and left sides of the partial area Z1 It is desirable to change smoothly between colors in the region. For this reason, the pixel values of the pixels constituting the lower part and the partial pixel values in the lower and left areas of the partial area Z1 are determined as sampling target pixel values. Specifically, the pixel value at the position indicated by “◯” in FIG. 7B is determined as a sampling target.

When the color approximation unit 3 determines the pixel value to be sampled for each partial region hierarchized by the region dividing unit 2, the color approximation unit 3 samples the pixel value to be sampled and approximates the pixel value with a continuous function ( Step ST9).
For example, as shown in FIG. 8, a local coordinate system (U, V) of a surface having point 1 as the origin (0, 0) is defined within a certain region. u and v are real parameters between “0” and “1”, the coordinates of point 2 are (u = 1, v = 0), the coordinates of point 3 are (u = 1, v = 1), The coordinates of 4 are defined as (u = 0, v = 1).
At this time, the color and luminance information of the point (u, v) sampled by the color approximating unit 3 can be approximated by an arbitrary parametric function F (u, v).

As the parametric function F (u, v), a Bezier curved surface or “Ferguson patch” described in Non-Patent Document 1 can be used.
Alternatively, a sigmoid function F (u, v) that can express a steep edge can be used.
F (u, v) = 1 / (1 + exp (a × u + b × v + c))
Here, a, b, and c are constants and may be arbitrarily specified by the user.

Alternatively, the pixel value may be approximated using a combination of basis functions as the parametric function F (u, v).
That is, the function F (u, v) may be configured using N basis functions f _i (u, v) (i = 1, 2,..., N).
F (u, v) = Σi f _i (u, v)
For example, “Radial basis function” can be used as the basis function.

When the image vectorization apparatus executes the processes of steps ST1 to ST9, vector image data is generated, and the vector image data is stored in the vector image data storage unit 4.
FIG. 9 is an explanatory diagram showing a configuration example of vector image data.
As shown in FIG. 9, the vector image data includes area information and an area access table. The area information is generated for each partial area belonging to each hierarchy, and the area access table is generated for each hierarchy.
For example, when there are M layers (1) to (M), M area access tables are generated.
As for the area information, for example, as shown in FIG. 4, there are 4 partial areas belonging to the hierarchy (1), 3 partial areas belonging to the hierarchy (2), and 6 partial areas belonging to the hierarchy (3). In some cases, four area information is generated in the hierarchy (1), three area information is generated in the hierarchy (2), and six area information is generated in the hierarchy (3).

Here, the area information is information composed of boundary line information, boundary line attribute information, color information, link area information, and internal line information.
The boundary line information is coordinate information generated by the area dividing unit 2, and this boundary line information includes a boundary line (a side surrounding the partial area, an oblique line or a curve included in the partial area). This is information indicating the position.
The boundary line attribute information is attribute information generated by the region dividing unit 2, and the boundary line attribute information is information indicating the attribute of the boundary line. For example, it indicates whether a side, a curve, or the like surrounding the partial region is a “continuous side” or a “discontinuous side”.

The color information is information relating to the color generated by the color approximating unit 3, and this color information is information indicating the type of continuous function approximating the pixel value of the partial area and the parameters of the continuous function. .
The link area information is information indicating the hierarchical relationship of the partial areas generated by the area dividing unit 2, and this link area information is an area for identifying a partial area of an upper hierarchy or a lower hierarchy linked to a certain partial area. This is information indicating a number, a hierarchy number indicating a hierarchy to which each partial area belongs, and the like.
The internal line information is information related to the internal line generated by the area dividing unit 2, and the internal line information includes the type of function indicating the oblique line or the approximate curve of the curve included in the partial area, and the function Information indicating parameters.

The area access table is a look-up table generated by the area dividing unit 2, and this area access table shows the correspondence between the image coordinates of the raster image and the partial areas hierarchized.
As will be described later, when an image coordinate indicating a certain viewpoint is given when drawing an image or the like of a certain viewpoint, a partial area including the image coordinates can be easily specified by referring to the area access table. Will be able to.
FIG. 10 is an explanatory diagram of a configuration example of the area access table.
Hereinafter, when an external device (for example, a car navigation device) draws an image of a certain viewpoint, the vector image data stored in the vector image data storage unit 4 is randomly accessed by referring to the area access table. A method for acquiring the pixel value will be described.

Here, for convenience of explanation, it is assumed that a 9 × 9 pixel raster image is divided into partial regions R1, R2, R3, and R4 as shown in FIG.
At this time, a minimum size table (region access table) that maintains the ratio of each partial region included in the original raster image is defined, and a region number indicating the partial region is substituted into the minimum size table.
In the example of FIG. 10B, the minimum size for maintaining the ratio of each partial area is 3 × 3 pixels (the size of the original raster image reduced to 1/3). Numbers 1 to 4 are assigned.

For example, consider a case where a partial region corresponding to the image coordinates (6, 5) is specified in the raster image of FIG.
In this case, the external device rounds the image coordinates (6, 5) to 1/3, which is the ratio of the area access table to the raster image, and calculates the rounded value (2, 2).
When the external device calculates the rounded value (2, 2), the rounded value (2, 2) is used as the coordinate value of the area access table, and the coordinates are set to the coordinates (2, 2) from the area access table shown in FIG. The assigned area number “4” is acquired.
When the external device acquires the region number “4”, the region information (for example, color information, internal line) in the partial region of the region number “4” from the vector image data stored in the vector image data storage unit 4 is acquired. Information, etc.) is read out, and the color of each pixel constituting the partial area (a part of the raster image including the image coordinates (6, 5)) is determined and drawn according to the area information.

As apparent from the above, according to the first embodiment, the edge detection unit 1 that detects an edge existing in a raster image and determines the importance of the edge in the raster image, and edge detection Among the edges detected by the section 1, the edge having the highest importance determined by the edge detection section 1 is used to divide the raster image into a plurality of partial areas, and the importance is lower than that edge. By using the edges in order and repeating the process of dividing the partial area into a plurality of partial areas, an area dividing unit 2 for hierarchizing the divided partial areas is provided, and the color approximating unit 3 is an area dividing unit For each partial area hierarchized by 2, the pixel value indicating the color of the pixels constituting the partial area is approximated by a continuous function. It is possible to dynamically change the used data amount of over data, an effect that may be random access to image data.

Hereinafter, the effect by this Embodiment 1 is demonstrated concretely.
First, since the region information constituting the vector image data generated by the image vectorization device is generated for each partial region that is hierarchized, an external device (for example, a car navigation device) can map at a certain scale or viewpoint. For example, area information necessary for drawing can be dynamically changed.

For example, consider a case where a vector image of a building is drawn on a three-dimensional map of a car navigation device.
When a bird's-eye view of a 3D map is used, the drawing viewpoint is away from the target object, so as shown in FIG. 11, using only lower layer region information (for example, region information of layer (1)), Display a rough image. At this time, it is sufficient to read only low-level area information on the memory of the car navigation apparatus.
On the other hand, when the drawing viewpoint approaches the target building on the 3D map, the details of the image are reproduced using also the high-level region information (for example, the region information of the layers (1) to (3)). To do.

As a result, only the necessary area information needs to be read in accordance with the image to be drawn. Therefore, when the amount of memory available in the car navigation device is small (generally, the amount of memory available for embedded hardware) However, it is possible to draw a desired image.
In addition, if an area access table constituting vector image data is used, random access to the vector image data is facilitated, and the pixel value of the designated coordinate is partially acquired without drawing the entire vector image. be able to.

In the present invention, any component of the embodiment can be modified or any component of the embodiment can be omitted within the scope of the invention.

The present invention is suitable for an image vectorization apparatus that needs to be able to draw a desired image even in a car navigation apparatus that has a small amount of available memory.

1 edge detection unit (edge detection means), 2 region division unit (region division unit), 3 color approximation unit (color approximation unit), 4 vector image data storage unit.

Claims

Edge detection means for detecting an edge present in the raster image and determining the importance of the edge in the raster image, and among the edges detected by the edge detection means, the edge detection means The raster image is divided into a plurality of partial areas by using the edge having the highest importance, and the partial areas are divided into a plurality of partial areas by sequentially using edges having a lower importance than the edges. Area division means for hierarchizing the divided partial area by repeating the division process, and a pixel indicating the color of the pixel constituting the partial area for each partial area hierarchized by the area division means An image vectorization apparatus comprising color approximation means for approximating a value with a continuous function.
The edge detection means reduces the raster image to a plurality of scales to detect edges existing in the images of a plurality of scales, and the edge detected from all the scale images has the highest importance. The edge detected only from a part of the scale images is determined to be an edge having a lower importance as an edge having a smaller scale image is detected. Image vectorization device.
2. The image vectorization apparatus according to claim 1, wherein the area dividing means obtains an approximated curve of a diagonal line or a curve included in the partial area after dividing the rectangular partial area.
The area dividing means assigns a hierarchy number indicating the hierarchy to which the divided partial area belongs to the partial area, and links the partial areas having a corresponding relationship among the partial areas belonging to different hierarchies. The image vectorization apparatus according to claim 1, wherein:
2. The image vectorization apparatus according to claim 1, wherein the area dividing means generates an area access table indicating a correspondence relationship between the image coordinates of the raster image and the partial areas hierarchized.
The color approximating means sets the attribute of the side surrounding the partial area divided by the area dividing means to the continuous side or the discontinuous side, and sets the attribute of the diagonal line or the curve included in the partial area to the discontinuous side. The image vectorization apparatus according to claim 1, wherein the image vectorization apparatus is set.
The color approximating means determines whether or not an edge exists on the side surrounding the partial area divided by the area dividing means, and sets the attribute of the side where the edge exists to a discontinuous side, while the edge does not exist The image vectorization apparatus according to claim 6, wherein the edge attribute is set to a continuous edge.
The color approximation means calculates a difference between the pixel value inside the partial area divided by the area dividing means and the pixel value of the peripheral area of the partial area, and if the difference between the pixel values is a predetermined value or more, The attribute of the side surrounding the partial region is set to a discontinuous side, while the attribute of the side surrounding the partial region is set to a continuous side if the difference between the pixel values is less than a predetermined value. Item 7. The image vectorization apparatus according to Item 6.
For the area surrounded by discontinuous sides, the color approximating means samples the pixel values in the area and approximates the pixel values with a continuous function, and the area surrounded by the continuous sides or For a region surrounded by diagonal lines or curves set to continuous sides and continuous sides, the pixel values in the region and a part of the pixel values in the adjacent region are sampled, and the pixel values are converted into a continuous function. The image vectorization apparatus according to claim 6, characterized by:
The edge detection means detects edges existing in the raster image, and the edge detection processing step for determining the importance of the edge in the raster image, and the area dividing means are detected in the edge detection processing step. Among the edges, the edge having the highest importance determined in the edge detection processing step is used to divide the raster image into a plurality of partial areas, and the edges having lower importance than the edges are used in order. Then, by repeating the process of dividing the partial area into a plurality of partial areas, an area division processing step for hierarchizing the divided partial areas, and a part in which the color approximation means is hierarchized in the area division processing step A color approximation processing step for approximating a pixel value indicating a color of a pixel constituting the partial area by a continuous function for each area. Method of.
An edge detection processing procedure for detecting an edge existing in the raster image and determining the importance of the edge in the raster image, and the edge detection processing among the edges detected in the edge detection processing procedure The raster image is divided into a plurality of partial areas using the edge having the highest importance determined in the procedure, and the partial areas are divided into a plurality of parts by sequentially using edges having a lower importance than the edges. Pixels constituting the partial area for each partial area hierarchized by the area division processing procedure for hierarchizing the divided partial area by repeating the process of dividing into partial areas An image vectorization program for causing a computer to execute a color approximation processing procedure for approximating a pixel value indicating a color of a color with a continuous function.