JP5504142B2 - Image processing apparatus, image processing method, and image processing program - Google Patents

Description

  The present invention relates to a technique for editing an area in an image represented by vector data using a parametric curved surface.

  In general, images taken with a digital camera, a video camera, or the like are often stored and used in a bitmap format. Bitmap data divides an image into pixels and handles all images as a set of pixels of various colors, so bitmap images are excellent in realism, but it is easy to edit images in units of objects. Instead, the image quality deteriorates when enlarged display is performed. Note that an object means a meaningful group captured in an image, and is represented by an image of a person, a head, a desk, a vehicle, or the like.

  On the other hand, vector-based data based on figures is easy to edit on an object-by-object basis because vector data is represented by functional numerical values and the object itself is drawn based on mathematical formulas. However, the accuracy of the object is maintained, and a constant image quality is maintained.

  Under such a background, a technique for converting a bitmap format image (hereinafter referred to as a bitmap image) into vector format data has been proposed (see Non-Patent Documents 1 and 2). These methods are characterized in that the original bitmap data is converted into vector data having as few graphic elements as possible without impairing the image quality (that is, the realism) of the original bitmap data as much as possible.

  Specifically, as shown in FIG. 13, an object in the captured bitmap image is regarded as a closed region, and the edge of the closed region and the inside thereof are changed based on a change in color in the closed region. Divide into “triangles composed of curves” and “rectangles composed of parametric curves”, approximate the contours of closed areas and boundaries of divided areas with parametric curves, and bit the colors inside the triangles or rectangles An approximation conversion to vector data is realized by referring to the colors of some of the pixels in the map image and interpolating based on mathematical expressions. Hereinafter, a triangle or rectangle whose sides are formed by parametric curves is referred to as a “curved surface patch”.

  The position of each vertex of the triangle or rectangle obtained by this conversion processing, the color value in each vertex or triangle, the control point of the parametric curve, etc. are saved as vector data in a file format or database format, and in the next processing By displaying an image using vector data, an object of the original bitmap image can be generated (reproduced) approximately. By this generation process, an image represented by vector data using a parametric curved surface is displayed. Therefore, the desired editing for the local area of the object in the image (object selection, enlargement or reduction of the selected area, Image processing such as vertex movement).

  In addition, when interpolating the color of the curved surface patch, the color information of the pixel position on the bitmap image corresponding to the position of the vertex constituting the curved surface patch or the appropriately sampled point is acquired, and the sampled point In general, the color values of points other than the position are approximately calculated using some mathematical formula. By calculating the color values inside the curved surface patch using mathematical formulas, even when the size of the object is enlarged, the color of the enlarged object can be expressed accurately, and deterioration in image quality can be prevented. .

  In other words, when the color value of each pixel of the bitmap image is retained as data as the color value inside the curved surface patch, a stepped boundary or color change appears in the contour portion during enlargement, and the image quality is impaired. However, by converting the geometrical shape of the area and the color information inside the area into vector format data (vector data) that can be calculated using mathematical formulas, smooth contours and color changes can be achieved even when enlarged. Can be expressed.

  Since the curved surface patch is easy to mathematically process, a bicubic patch whose side is represented by a parametric curve of degree 3 is often used. Normally, this bicubic patch is used to represent a three-dimensional curved surface shape. However, in Non-Patent Documents 1 and 2, a process of forming a curved patch shape along a contour of an object from a bitmap image, This is used to calculate the color value inside the curved surface patch.

  For example, in Non-Patent Documents 1 and 2, the side of the curved patch uses a cubic Bezier curve or a curve equivalent to the curve. In Non-Patent Document 1, (x, y, h) is represented by a Ferguson patch when the pixel position (x, y) and the color value h of the pixel position are variables. The patch shape (x, y) is represented by a Bezier curve, and the color value at the pixel position (x, y) is calculated using a thin plate spline interpolation method.

  In general, a parametric curve has a parameter of the position in the horizontal axis direction of the screen as u (real number of 0 ≦ u ≦ 1) and a parameter of the position in the vertical axis direction as v (real number of 0 ≦ v ≦ 1) (u , V) is represented by vector data having a position vector at a point determined by (x, y, h). Note that x, y, and h are vector components of the local coordinate system of the x-axis, y-axis, and h-axis, respectively, and (x, y) means the position of the point on the curved surface patch, h Means the color value at (x, y).

Then, the parameter, u = (u, v) _{and, u i = (u i,} v i) marked with subscripts expressed in such, x, y, h is follows by using the expression form It calculates from Formula (1)-Formula (3).

x _i = x (u _i , v _i ) (1)
y _i = y (u _i , v _i ) (2)
h _i = h (u _i , v _i ) (3)
Although it seems that the positions corresponding to the parameters u and v are calculated first on these equations, when the vector data is actually rendered as an image (color value calculation, etc.) and displayed, After specifying the pixel position (x, y) and calculating the value of the parameter (u, v) at that position, the color value h (u, v) corresponding to the calculated (u, v) is calculated. Will do.

  For this reason, the method of displaying a screen using vector data as disclosed in Non-Patent Documents 1 and 2 is computationally expensive compared to displaying a bitmap image having the same size (resolution).

  In view of this, a method is used in which an identifier is assigned to each of a plurality of curved patches approximately applied to an object, and the unit for editing is limited to the specified object, thereby reducing the calculation cost.

Yu-Kun Lai, two others, “Automatic and Topology-Preserving Gradient Mesh Generation for Image Vectorization”, ACM Transactions on Graphics, Vol.28, No.3 (SIGGRAPH 2009), August 2009, Article 85: 1- 8 Tian Xia, two others, “Patch-based Image Vectorization with Automatic Curvilinear Feature Alignment”, ACM Transactions on Graphics, Vol.28, No.5 (SIGGRAPH Asia 2009), December 2009, Article 115: 1-10

  However, since the curved surface patch used is composed of a parametric curved surface formed by a curve (parametric curve), the shape of the curved surface patch and the color value inside the curved surface patch are used when the object in the image is deformed. The recalculation requires a lot of processing time, and there is a problem that the display speed of the image after the deformation is slow.

  The present invention has been made in view of the above, and an object of the present invention is to display and edit a vectorized image at high speed without losing image quality as much as possible.

The image processing apparatus according to claim 1, storage means for storing vector data of a rectangular or triangular curved surface patch obtained by approximating an object in a bitmap image with a parametric curved surface, and reading the vector data from the storage means. Generating means for generating a primitive patch connecting the vertices constituting the curved surface patch with line segments; dividing a curved edge connecting two vertices of the curved surface patch; dividing points on the divided curved edge and the curved surface; Dividing means for dividing the primitive patch by a line segment connecting to other vertices of the patch, array data corresponding to each pixel of the bitmap image is generated, and at a position corresponding to the inner area of the divided primitive patch Storage means for storing an identifier for identifying a primitive patch, and an identifier corresponding to a designated pixel position in the array Obtained from over data, characterized by having a process and processing means for executing in response to a request using the identifier of the acquired primitive patch.

  According to the present invention, processing is performed using an identifier of a primitive patch having a straight boundary instead of a curved surface patch whose side is composed of a parametric curve, so that the calculation cost related to rendering is reduced and high speed is achieved without losing image quality as much as possible. The vectorized image can be displayed and edited.

The image processing device according to claim 2 is the image processing device according to claim 1, wherein the color value of the pixel corresponding to the primitive patch after the vertex movement is changed to the vertex of the primitive and the primitive image updated by the vertex movement. And determining means for determining the position using the position information of the sampling point, the vertex and the color value before the sampling point is updated.

The image processing method according to claim 3 is the image processing method performed in a computer, the storage unit may store the object in the bitmap image in the storage unit vector data surface patch of rectangular or triangular is approximated by a parametric surface A step of generating vector data from the storage unit, generating a primitive patch connecting vertices constituting the curved patch with line segments, and a dividing unit including two of the curved patch dividing the curved edges connecting the vertices, dividing the primitive patch line connecting the other vertexes of division points and the surface patch on the divided curved edges, the storage unit, of said bitmap image Array data corresponding to each pixel is generated, and the data is plotted at a position corresponding to the inner area of the divided primitive patch. And storing an identifier for identifying the Mitibupatchi, processing means, and obtains an identifier corresponding to the designated pixel positions from the sequence data, in response to a request by using the identifier of the acquired said primitive patching And performing .

  According to the present invention, processing is performed using an identifier of a primitive patch having a straight boundary instead of a curved surface patch whose side is composed of a parametric curve, so that the calculation cost related to rendering is reduced and high speed is achieved without losing image quality as much as possible. The vectorized image can be displayed and edited.

An image processing method according to a fourth aspect is the image processing method according to the third aspect, wherein the determining means sets the color value of the pixel corresponding to the primitive patch after the vertex movement of the primitive updated by the vertex movement. The method further comprises the step of determining using the position information of the vertices and the sampling points in the primitive image and the color values before the vertices and the sampling points are updated.

  According to a fifth aspect of the present invention, there is provided an image processing program that causes a computer to execute each step in the image processing method according to the third or fourth aspect.

  According to the present invention, it is possible to display and edit a vectorized image at high speed without losing image quality as much as possible.

It is a figure which shows the functional block structure of an image processing apparatus. It is a figure which shows the operation | movement flow of an image processing apparatus. It is a figure which shows the example of approximation of a bitmap image, and the example of label data. It is a figure which shows the vector data example. It is a figure which shows the vector data example. It is a figure which shows the example which approximates a curve edge with a line segment. It is a figure which shows the correlation of a curved surface patch and a parametric patch. It is a figure which shows the example of a division | segmentation of a curve edge. It is a figure which shows the example of a division | segmentation of a curve edge and a primitive patch. It is a figure which shows the vertex movement example of a primitive patch. It is a figure which shows the correction example of the curve of the primitive patch after moving a vertex, a vertex, and a control point. It is a figure which shows the example of correction of the curve, vertex, and control point of the divided | segmented primitive patch after vertex movement. It is a figure which shows the example of an approximation in a parametric curved surface.

  Hereinafter, an embodiment for carrying out the present invention will be described using a triangular curved patch and a triangular primitive patch (described later) as an example. However, the present invention can be implemented in many different modes and should not be construed as being limited to the description of the present embodiment.

  That is, even when a rectangular curved surface patch or a rectangular primitive patch is used, the same effect can be obtained. Hereinafter, a triangular or rectangular curved patch is simply referred to as a curved patch, and a triangular or rectangular primitive patch is simply referred to as a primitive patch.

  The parametric curve constituting one side of the curved patch is represented by a Bezier curve (Bezier curve) or a rational Bezier curve, but the theorem for making the one side a curve is not limited at all.

  FIG. 1 is a diagram showing a functional block configuration of the image processing apparatus according to the present embodiment. The image processing apparatus includes a data storage unit 11, a vector data reading / display unit 12, an identifier assigning unit 13, a first label data generation / storage unit 14, an identifier acquisition unit 15, an image processing unit 16, Vector data update unit 17, primitive patch generation unit 18, curve edge / line segment edge evaluation unit 19, parametric curve division unit 20, primitive image generation / display unit 21, and second label data generation / storage unit 22 A vertex movement receiving unit 23, a primitive image update / display unit 24, and a second label data update / storage unit 25.

  The data storage unit 11 has a function of storing vector data constituting a curved surface patch obtained by approximating an object in the original bitmap image with a parametric curved surface.

  The vector data reading / display unit 12 reads the vector data in the data storage unit 11, generates a curved surface patch using the vector data, and generates a bitmap image (raster image) based on the generated curved surface patch. And has a function of displaying on the display. That is, it has a function of approximating the original bitmap image from the vector data digitized / functioned based on the original bitmap image.

  The identifier assigning unit 13 applies each curved surface patch to the curved surface patch generated during the bitmap image generation in the vector data reading / displaying unit 12 and the sides constituting the curved surface patch (hereinafter, curved edges). It has a function of assigning an identifier for identifying each curved edge.

  The first label data generation / storage unit 14 is two-dimensional array data (hereinafter referred to as first label data) having the same size as the original bitmap image and having a data area corresponding to each pixel of the bitmap image. And the identifier of the curved surface patch assigned by the identifier assigning unit 13 is stored (stored) in each position corresponding to the internal area of the curved patch.

  The identifier acquisition unit 15 has a function of accepting an edit processing request by a user and acquiring an identifier of an object patch corresponding to the pixel range designated by the request from first label data or second label data (described later). doing.

  The image processing unit 16 has a function of performing predetermined processing (image processing such as object selection, enlargement / reduction of a selection area, vertex movement, etc.) according to a request using the patch identifier acquired by the identifier acquisition unit 15. doing.

  The vector data update unit 17 has a function of updating (correcting, adding, etc.) the vector data in the data storage unit 11 using the vector data corrected by the processing of the image processing unit 16.

  The primitive patch generation unit 18 has a function of generating a primitive patch in which vertices constituting a curved surface patch are connected by line segments.

  The curved edge / line segment edge evaluation unit 19 determines the difference in the distance (the edge component length) between the curved edge connecting two vertices of the curved patch and the side of the primitive patch corresponding to the curved edge (hereinafter, the line segment edge). It has a function to evaluate (difference).

  The parametric curve dividing unit 20 divides the curved edge of the curved surface patch based on the evaluation of the difference in distance by the curved edge / line segment edge evaluating unit 19, and adds the dividing points on the divided curved edge and the curved surface patch. It has a function of dividing a primitive patch by a line segment connecting the vertices.

  The primitive image generation / display unit 21 obtains the color value of the pixel position corresponding to the internal region of the primitive patch (including the divided primitive patch) from the bitmap image based on the curved surface patch generated by the vector data reading / display unit 12. And a primitive image based on the primitive patch is generated and displayed on the display.

  The second label data generation / storage unit 22 generates second label data having the same area as the first label data generation / storage unit 14 and corresponds to the internal area of the primitive patch (including the divided primitive patch). It has a function of storing (storing) an identifier for identifying a primitive patch at a position to be stored.

  The vertex movement accepting unit 23 has a function of accepting designation of a vertex of a primitive patch to be moved on the primitive image and a movement amount thereof.

  The primitive image update / display unit 24 has a function of updating the primitive image based on the position and amount of movement of the primitive patch after the vertex movement and displaying the primitive image on the display.

  The second label data update / storage unit 25 has a function of updating (correcting) the identifier of the primitive patch in the second label data in accordance with the update of the primitive image.

  Next, the operation of the image processing apparatus according to the present embodiment will be described with reference to FIG.

  First, a processing method based on a curved surface patch will be described. The main feature of the present invention is a processing method based on primitive patches based on line segments, which will be described later. Since the processing method uses an image generated based on a curved surface patch, the processing method of the curved surface patch is used. I will explain it first.

  First, the vector data reading / display unit 12 reads the vector data stored in the data storage unit 11, generates the shape of the curved surface patch using the vector data, and interpolates the color value inside the curved surface patch. A bitmap image with a specified image size is generated (S101, S103). See FIG. 3A for an example of the generated bitmap image.

  Note that since a method for generating a bitmap image using vector data is described in Non-Patent Document 2, description thereof is omitted.

  The reason why the image size is specified is that the vector data does not depend on the size of the bitmap image to be generated, and therefore the image size is required to generate the image. In general, the horizontal direction is a pixel of W (a natural number of 1 or more) and the vertical direction is a pixel of H (a natural number of 1 or more).

  Here, the vector data will be described. The vector data here is data that is a condition for determining a function for constructing a curved surface patch. For example, as shown in FIG. 4, the vertex IDs, vertex coordinate positions, and vertexes constituting the curved surface patch are used. A data table composed of color values, etc., a data table composed of control points for controlling the curved edges of curved patches, sampling points used for color interpolation, and the like. The format and format of the vector data can be arbitrarily set, and there is no restriction on the method.

  Returning to FIG. 2, when generating a bitmap image, the identifier assigning unit 13 assigns identifiers to the plurality of generated curved patches and the plurality of curved edges of each curved patch (S102).

  In addition, as an identifier given in this way, a value obtained as appropriate may be assigned, or when an identifier is included in vector data, the identifier may be used.

  Next, when the editing process by the user is only patch selection, the first label data generation / storage unit 14 generates first label data corresponding to the original bitmap image and is given by the identifier giving unit 13 The identifier of the curved patch is stored in each position corresponding to the internal area of the curved patch (S104, S105). Refer to FIG. 3B for an example of the generated first label data.

  Specifically, the position (i, j) (0 ≦ i <W, 0 ≦ j <H) corresponding to the image occupied by each curved surface patch and the color value at that position are calculated. The first label data is generated by substituting T for all the calculated positions (i, j) for the curved surface patch to which the identifier is assigned.

  Next, the vector data reading / display unit 12 displays the bitmap image generated in S103 on the display (S106).

  Thereafter, the identifier acquisition unit 15 accepts a curved patch selection process for the bitmap image displayed on the display, and the identifier of the curved patch corresponding to the pixel range specified by the selection is generated from the first label data generated in S105. The acquired image processing unit 16 performs processing using the acquired curved surface patch identifier (S107 to S109).

  By acquiring the identifier of the curved surface patch in S108, for example, a plurality of curved surface patches can be collectively handled as one object, and some tag information can be given. By adding the tag information to the read vector data, new information can be added to the vector data.

  For example, by adding a list such as (tag ID, patch 1 ID, patch 2 ID, patch 3 ID,...) To the vector data, a patch ID having the tag ID can be acquired at once. It becomes possible.

  Finally, the vector data update unit 17 updates the vector data in the data storage unit 11 using the vector data updated in S109 (S110).

  Note that S107 to S110 are repeated until the curved patch selection processing is completed (S111).

  Further, in order to show the selected curved surface patch in an easy-to-understand manner, the color of the selected curved surface patch on the bitmap image may be changed. For example, this is realized by setting the color value of the pixel position corresponding to the position having the same identifier on the first label data to the same color.

  The above is the basic processing method based on the curved surface patch. The processing method described so far may be used as long as it is only processing that handles identifiers such as patch selection, but in the case of deforming an object in a bitmap image, that is, when moving a vertex constituting a curved surface patch, A lot of calculation costs are required to recalculate the shape of the curved patch after movement and the internal color values.

  Therefore, in this case, processing is performed using a primitive patch in the form of a primitive (polygon: a triangle such as a triangle having a line segment side or a rectangle) having a line segment edge that can generally be processed at high speed. Hereinafter, the processing method based on the primitive patch will be described with reference to FIG.

  In the determination process in S104, if the editing process by the user is not only the patch selection, it is determined whether there is a curve edge that has not yet been processed among the curve edges of the selected curved patch, and an unprocessed curve edge is determined. If there is, the vector data reading / display unit 12 and the identifier assigning unit 13 use the vector data in the data storage unit 11 to generate an information list regarding curved patches and curved edges (S104, S201 to S202). See FIG. 5 for an example of the generated list. Such a list may be stored in the data storage unit 11 in advance.

  The parent patch ID shown in FIG. 5 is an identifier indicating to which patch a patch newly generated at the time of division described later belongs. Similarly, the parent edge ID is an identifier indicating to which edge the edge newly generated at the time of division described later belongs.

  Next, returning to FIG. 2, the primitive patch generation unit 18 generates a primitive patch in which the vertexes of the curved surface patch are connected by line segments using the vector data in each list generated in S202 (S203).

  Here, as a method of generating a primitive patch of a line segment from a curved patch of a parametric curved surface, a method of generating a polygon by connecting a contour as a parametric curve and connecting only the inside with a line segment as shown in FIG. A method of generating a polygon by connecting the contours as shown in FIG. For generating the label data, the identifier is stored with reference to the polygon, so that the recalculation cost such as the color value can be reduced even if the vertex is moved by any method.

  Here, as shown in FIG. 7, even if the vertex positions are the same, the identifier of the designated pixel position may be different between the curved surface patch and the primitive patch represented by the original vector data.

  Returning to FIG. 2, the curved edge / line segment edge evaluation unit 19 calculates the difference between the curved distance of the curved edge constituting the curved surface patch and the linear distance of the line segment edge of the primitive patch corresponding to the curved edge. If the difference in distance is greater than the predetermined threshold, the parametric curve dividing unit 20 divides the curve edge at the midpoint (S204 to S206).

  Here, a distance difference evaluation method will be described with reference to FIG. For example, it can be evaluated by a chessboard distance between two points. The chessboard distance d between the two vertices v1 (x1, y1) and v2 (x2, y2) can be defined as the following equation (4).

d (v1, v2) = max (| x1-x2 |, | y1-y2 |) Formula (4)
When the two vertices v1 and v2 and the two control points c1 and c2 constituting the curve are arranged in the order of v1, c1, c2, and v2, the curve distance of the curve edge is calculated by the equation (5). The straight line distance of the line segment edge is calculated by equation (6).

d1 = d (v1, c1) + d (c1, c2) + d (c2, v2) (5)
d2 = d (v1, v2) (6)
Thereafter, when the difference between d1 and d2 is larger than a predetermined threshold value, the curve edge is divided.

  In addition, when the curved edge is represented by the position (u, v = 0) in the expressions (1) to (3), the curve edge is divided at the position (u = 1/2, v = 0). May be.

  As for the method of dividing the parametric curve in this way, that is, the method of increasing the number of segments while maintaining the curve shape, “Curves and Surfaces for CAGD; 5th edition” (Gerald Farin, Academic Press, 2002) (below) This is a well-known technique as shown in Reference 1).

  Here, dividing a curved edge affects two patches that have the curved edge as a boundary. Therefore, before generating a primitive patch, determine whether to divide all curved edges and determine that they are to be divided. The curved edge may be divided in advance.

  Next, returning to FIG. 2, the parametric curve dividing unit 20 generates control points for each curve edge divided by using the dividing method described in Reference Document 1 above, and assigns a new ID for each control point. Assigned and added to the data table in association with the ID of the curve edge before division (S206).

  Next, the parametric curve dividing unit 20 connects the dividing points on the divided curve edges and the other vertices of the two primitive patches (the vertices that are not the two vertices constituting the curve edge) having the curve edge as a boundary. Add a new edge, divide each of the two primitive patches by the added new edge, assign a new ID to the divided primitive patch or edge, and associate it with the ID of the curve edge before the division Are added to the data table (S206 to S208).

  Refer to FIG. 9 for the dividing procedure of S206. The newly added edge does not need to be associated with a curve in particular, and may not have a control point.

  Here, care must be taken when dividing the two edges constituting one patch. When vertexes v1, v2, and v3 of a patch have edges e1 = (v2, v3), e2 = (v3, v1), e3 = (v1, v2), the reverse of when e2 is divided and e2 is divided The division result may differ from the case where e1 is divided after dividing e2. Therefore, when only a specific primitive patch is considered, it is possible to preferentially divide the edge that constitutes the patch from the edge having a large distance difference so that the division becomes unique. . Alternatively, the distances of the edges of all patches may be evaluated in advance, and the patches may be divided in descending order of the distance difference, and the primitive patches may be divided after the edge division is completed.

  In addition, when the difference in distance also exceeds a predetermined threshold value for the edge after division, the division of the divided edge may be repeated until the difference in distance becomes equal to or less than the predetermined threshold value. The number of divisions may be determined in advance.

  Next, the primitive image generation / display unit 21 acquires the color value of the pixel position corresponding to the internal area of the primitive patch generated in S203 and S206 from the bitmap image based on the curved surface patch generated in S103, and the primitive A primitive image based on the patch is generated (S209). Refer to FIG. 3C for an example of the generated primitive image.

  As the primitive image, the bitmap image generated in S103 may be used as it is, or a duplicate of the bitmap image may be used.

  Next, the second label data generation / storage unit 22 generates the second label data corresponding to the primitive image, and sets the identifier of the primitive patch to each position corresponding to the internal area of the primitive patch (including the divided primitive patch). (S210). Refer to FIG. 3D for an example of the generated second label data.

  Next, the primitive image generation / display unit 21 displays the primitive image generated in S209 on the display (S211).

  Thereafter, the identifier acquisition unit 15 accepts a primitive patch selection process for the primitive image displayed on the display, and acquires the identifier of the primitive patch corresponding to the pixel range designated by the selection from the second label data generated in S210. Then, the image processing unit 16 performs processing using the acquired primitive patch identifier (S212 to S214).

  Finally, the vector data update unit 17 updates the vector data in the data storage unit 11 using the vector data updated in S214 (S215).

  Note that S212 to S215 are repeated until the primitive patch selection processing is completed (S216).

  Further, since it is label data for a primitive patch that approximates a curved surface patch with a primitive, as described with reference to FIG. 7, there is a possibility that an adjacent identifier is acquired in the vicinity of an edge instead of an accurate identifier. The label expression by the primitive edge has an effect that the vertex movement described later can be realized at high speed.

  The operation when the vertex of the primitive patch moves will be described with reference to FIG.

  After S213, the vertex movement receiving unit 23 receives the designation of the vertex coordinates to be moved and the movement amount on the primitive image displayed in S211 (S301).

  Next, the primitive image update / display unit 24 modifies the primitive image based on the moved position and movement amount received in S301 (S302).

  As shown in FIG. 10, the vertex movement of the primitive patch affects a plurality of primitive patches sharing the vertex, but the color value in the primitive patch can be easily changed even if the shape of the primitive patch is deformed by the vertex movement. It can be calculated.

  For example, with respect to the movement of the vertex, by using the affine transformation shown in Reference Document 1 above, it is possible to easily calculate from which position before the movement the coordinate value inside the moved primitive patch is moved. it can. The color value can be easily calculated by using the color value at the position before the movement.

  At the same time as S302, the second label data update / storage unit 25 corrects the identifier of the primitive patch in the second label data in accordance with the update in S302 (S303). Only the primitive patch including the vertex to be moved can be realized at low cost because the identifier is corrected.

  Finally, proceeding to S215, the vector data updating unit 17 updates the vector data in the data storage unit 11 based on the moved position and movement amount received in S301.

  Here, an example of updating vector data when there is a vertex movement will be described with reference to FIG. 11 or FIG. FIG. 11 shows a vertex and a curved edge related to the vertex. Even if the curve edge is approximated by a line segment edge, the curve can be easily restored if there is a control point.

  In FIG. 11, it is assumed that the line segment edge of e1 is moved to e1a by moving the vertex of v2 to v2a. At this time, the vertices v1 and v2 and the control points c1 and c2 for forming the original curved edge are also affected, but in the present invention, only the control point c2 close to v2 and v2 is moved, and the movement target Only update the vector data. On the other hand, when c2 is not determined in position, it is not necessary to move c2. However, for example, the coordinate value of c2a using a constraint condition that the Euclidean distance between v1 and c2a and the Euclidean distance between v1 and c2a do not change. You may decide.

  FIG. 12 shows a case where the vertex of the divided primitive patch moves. Even in this case, since the curved edges are smoothly connected at the dividing points, only the control point closer to the moved vertex is moved and the vector data of only the moving target is updated as described above. To do. Thereby, the smooth connection of the edge can be maintained at the dividing point.

  That is, in S215 of FIG. 2, the vector data update unit 17 updates the vector data of only the vertex to be moved and the control point close to the vertex for the patch whose edge is not divided.

  If necessary, an appropriate sampling point is extracted from the updated primitive image, and the point coordinates necessary for patch interpolation and the color value at the point coordinates are acquired to correspond to the primitive patch after the vertex movement. The color value of the pixel may be determined and added to the vector data. For example, the color value of the pixel corresponding to the primitive patch after moving the vertex, the position information of the vertex of the primitive updated by moving the vertex and the sampling point in the primitive image, and the color value before updating the vertex and sampling point And add to the vector data.

  On the other hand, for a patch whose edge is divided, for the patch increased by the division, the vertex, the control point, the sampling point, and the color value of each point constituting the patch are added to the vector data. In addition, after vertexes and control points so that two or more curved edges of the increased two or more patches can be represented by one parametric curve, the vertex, control points, sampling points, and color values of each point are vector data. You may add to each.

  If necessary, an appropriate sampling point is extracted from the updated primitive image, and the point coordinates necessary for patch interpolation and the color value at the point coordinates are acquired to correspond to the primitive patch after the vertex movement. The color value of the pixel may be determined and added to the vector data. For example, the color value of the pixel corresponding to the primitive patch after moving the vertex, the position information of the vertex of the primitive updated by moving the vertex and the sampling point in the primitive image, and the color value before updating the vertex and sampling point And add to the vector data.

  The method of approximating a plurality of connected curves with a single parametric curve is, for example, to select vertices constituting a plurality of curves and appropriate points on the curves, and pass the vertices and appropriate points. It is also possible to calculate the control point position where a simple parametric curve can be obtained. Such an approximation method is called data fitting, and is shown, for example, in “An Algorithm for Automatically Fitting Digitized Curves” (Philip J. Schneider, Graphics Gems, Morgan Kaufmann Publishers, pp.612-626, 1990.). It is a well-known technique.

  According to the present embodiment, processing is performed using an identifier of a primitive patch having a straight boundary instead of a curved surface patch whose edge is formed by a parametric curve, so that the calculation cost for rendering is reduced and the image quality is impaired as much as possible. It is possible to display and edit a vectorized image at high speed. Further, the editing work can be reflected in the original vector data based on the high-speed display.

  Hereinafter, the action on the effect obtained by using the identifier of the primitive patch having the straight line boundary will be described in detail.

  An image composed of curved patches with sides composed of parametric curves (this is equivalent to dividing the original bitmap image with curved patches composed of sides with parametric curves) When rendering (drawing), the color value of the position corresponding to each pixel on the bitmap screen is determined by calculation, and it is determined to which curved surface patch this “position corresponding to each pixel” belongs. Is not easy.

  This is because even if one pixel is selected, even if it is determined whether it is inside or outside one curved patch, a lot of calculations are required. For example, in order to determine whether or not a pixel at a certain position (x, y) is included in the curved surface patch P1, the values of (u, v) from the above formulas (1) and (2) are set to, for example, Newton-Raphson. Generally, it is generally determined by numerical calculation using a method or the like, and it is determined by evaluating whether it is 0 ≦ u ≦ 1 or 0 ≦ v ≦ 1.

  However, if it is not included in the curved surface patch P1, this determination is repeated until a curved surface patch including the pixel is found, for example, the same calculation is performed for another curved surface patch P2.

  For the same reason, when a user selects a certain pixel on an image composed of curved patches, it is not easy to determine which curved patch is inside.

  Therefore, when rendering a vectorized image with a curved patch, the identifier of the curved patch is written into the first label data at the same time, and when a pixel is selected after image generation, numerical analysis using the above formula is performed. By determining based on the identifier stored at the position on the first label data, instead of determining whether the surface patch is based on the inside or outside, it is possible to immediately specify the curved surface patch including the pixel position.

  However, in this method, when editing such as deforming a curved surface patch is performed, it takes time for the above reason to render a new image after deformation, so it cannot be said that the method is efficient.

  On the other hand, a primitive patch whose sides are composed of straight lines (line segments) can determine whether a certain point is inside or outside by relatively simple calculation. Specifically, it is only necessary to simply calculate whether or not the image is included in a triangle, and in the case of a rectangle, the same calculation may be performed twice by dividing the triangle into two triangles. Furthermore, if the identifier stored in the second label data is used for the inside / outside determination, even such a calculation becomes unnecessary. In addition, regarding the recalculation of the color after the deformation of the primitive patch and the update of the second label data, it is easy to determine the inside / outside of the pixel.

  Therefore, it is possible to reduce the calculation cost for rendering, and it is possible to display and edit a vectorized image at high speed without losing image quality as much as possible.

  Finally, the image processing apparatus described in this embodiment is configured by a computer. That is, each processing function unit constituting the image processing apparatus is realized by a storage unit such as a memory or a hard disk, or an arithmetic unit such as a CPU, and is executed by a program.

  In addition, the image processing apparatus described in this embodiment is recorded as a program in a recording medium such as an optical storage device or a magnetic storage device in a readable manner, and the recording medium is incorporated in a computer or recorded on the recording medium. Can be downloaded to a computer via an arbitrary communication line, or installed from a recording medium, and the computer can be operated by the program so that each operation described above can function as an image processing apparatus. is there.

DESCRIPTION OF SYMBOLS 11 ... Data memory | storage part 12 ... Vector data reading / display part 13 ... Identifier provision part 14 ... 1st label data generation / storage part 15 ... Identifier acquisition part 16 ... Image processing part 17 ... Vector data update part 18 ... Primitive patch generation part DESCRIPTION OF SYMBOLS 19 ... Curve edge / line segment edge evaluation part 20 ... Parametric curve division part 21 ... Primitive image generation / display part 22 ... Second label data generation / storage part 23 ... Vertex movement acceptance part 24 ... Primitive image update / display part 25 ... Second label data update / storage unit S101-S111, S201-S216, S301-S303 ... step

Claims (5)

Storage means for storing vector data of rectangular or triangular curved surface patches obtained by approximating an object in a bitmap image by a parametric curved surface;
Generating means for reading vector data from the storage means and generating a primitive patch connecting the vertices constituting the curved surface patch with line segments;
A dividing unit that divides a curved edge connecting two vertices of the curved surface patch, and divides the primitive patch by a line segment connecting a dividing point on the divided curved edge and the other vertex of the curved surface patch;
Storage means for generating array data corresponding to each pixel of the bitmap image and storing an identifier for identifying the primitive patch at a position corresponding to the inner region of the divided primitive patch;
Processing means for acquiring an identifier corresponding to a designated pixel position from the array data, and performing processing according to a request using the acquired identifier of the primitive patch;
An image processing apparatus comprising: The color value of the pixel corresponding to the primitive patch after the vertex movement, the position information of the vertex of the primitive updated by the vertex movement and the sampling point in the primitive image, and the color value before the update of the vertex and the sampling point The image processing apparatus according to claim 1, further comprising a determination unit configured to determine using An image processing method performed in a computer,
A step of storing means stores the object in the bitmap image in the storage unit vector data surface patch of rectangular or triangular is approximated by a parametric surface,
A generation unit that reads vector data from the storage unit and generates a primitive patch connecting vertices constituting the curved surface patch with line segments;
Dividing means for dividing a curved edge connecting two vertices of the curved surface patch, and dividing the primitive patch by a line segment connecting a dividing point on the divided curved edge and another vertex of the curved surface patch; ,
Storing means for generating array data corresponding to each pixel of the bitmap image and storing an identifier for identifying the primitive patch at a position corresponding to the inner region of the divided primitive patch;
A processing unit that acquires an identifier corresponding to a designated pixel position from the array data, and performs processing according to a request using the acquired identifier of the primitive patch;
An image processing method comprising: The determination means determines the color value of the pixel corresponding to the primitive patch after the vertex movement , the position information of the vertex of the primitive updated by the vertex movement and the sampling point in the primitive image, and before the update of the vertex and the sampling point. The image processing method according to claim 3, further comprising a step of determining using the color values of the image data.   An image processing program for causing a computer to execute each step in the image processing method according to claim 3 or 4.