JP5620313B2 - Image processing apparatus, method, and program - Google Patents

Description

  Embodiments described herein relate generally to an image processing apparatus, a method, and a program.

  An image that generates a thumbnail image of the original image by detecting one object region (building, person, etc.) in the original image to be processed and distorting the other region while enlarging the object region There is a processing device.

  In a conventional image processing apparatus, when there are a plurality of object areas, if one object area is enlarged, the other object areas may be distorted, and it is possible to generate a thumbnail image that makes it easy for the user to specify the original image. It was difficult.

  Such an image processing apparatus is desired to be able to generate a thumbnail image that allows a user to easily specify an original image.

JP 2006-313511 A Japanese Patent No. 3482923

Norihiro Nakamura, Hidehiko Abe, Koji Nishio, Kenichi Kobori, “A method for resolution conversion using Delaunay triangulation”, Journal of the Institute of Image Electronics Engineers of Japan, Vol.35, No.5, 2005. M. Kass, A. Witkin, and D. Terzopoulos, "Snakes: Active Contour Models." International Journal of Computer Vision, Vol.1, No.4, pp. 321-331, 1988.

  The problem to be solved by the present invention is to provide an image processing apparatus, method, and program capable of generating a thumbnail image that allows a user to easily specify an original image.

  In order to solve the above problems, an image processing apparatus according to an embodiment of the present invention includes a detection unit, a setting unit, a calculation unit, and a deformation unit.

  The detection unit detects the first object region and the second object region from the original image to be processed. The setting unit sets, from the original image, a deformation area that includes the first object area and does not include the second object area. The calculation unit searches for an enlargement position where the size of the first object area can be enlarged in the deformation area, and calculates an enlargement magnification at the enlargement position. The deforming unit enlarges the first object area at the searched enlargement position at the enlargement magnification and performs a deformation process on the remaining part of the deformation area, and performs a deformation process on the second object area. The deformed original image is generated by reducing the deformation process.

Explanatory drawing of the image processing by the image processing apparatus 1 which concerns on 1st Embodiment. Explanatory drawing of an original image. 1 is a block diagram illustrating an image processing apparatus 1. FIG. 4 is a flowchart showing processing of the image processing apparatus 1. 7 is a flowchart showing processing of the calculation unit 14. Explanatory drawing of the process of the calculation part. 10 is a flowchart showing a part of processing of a calculation unit 14; An example figure showing candidate point P'm. Explanatory drawing of the process of the classification | category part 151. FIG. The flowchart showing the process of the area | region deformation | transformation part 152. FIG. Explanatory drawing of the deformation | transformation process of the area | region 305. FIG. Explanatory drawing of the deformation | transformation process of the area | region 306. FIG. 5 is a flowchart showing enlargement processing of a target object area 301. Explanatory drawing of the expansion process of the target object area | region 301. FIG. The flowchart showing the process of the setting part 13 in 2nd Embodiment. An example figure of a mesh.

(First embodiment)
The image processing apparatus 1 according to the first embodiment is for generating a thumbnail image of an original image from the original image. The image processing apparatus 1 can be used for an apparatus capable of displaying an image, such as a television, a PC (personal computer), a portable information terminal, and a digital camera. The image processing apparatus 1 sets a deformation area that includes a target object that the user pays attention to in the original image and does not include other object areas, and generates a deformation image in which the target object is moved and enlarged in the deformation area. . A thumbnail image is obtained by reducing the deformed image.

  FIG. 1 is an explanatory diagram of image processing by the image processing apparatus 1. The image processing apparatus 1 moves and enlarges the target object region 301 to be processed in the original image to a position where the size of the target object region 301 becomes the largest in the deformation region 302 (FIG. 1 (a) to FIG. 1 (c). ) Shows the process). The deformation area 302 is a partial area of the original image and is an area where the movement and enlargement of the object 301 are permitted.

  FIG. 2 is an explanatory diagram of an original image in the present embodiment. An image having pixel values distributed two-dimensionally can be expressed as a set (mesh) of a large number of small regions. Here, the mesh is expressed by a triangle (patch) made up of a plurality of vertices and ridge lines (edges) connecting the vertices. The patch may be a polygon other than a triangle.

  One vertex is position (x, y) and connection information indicating which vertex is connected to which vertex (for example, “vertex P1 is connected to vertex P2 to vertex P8 in FIG. 2”). Information). Each patch may be assigned a patch ID for identifying each patch. Each vertex in the mesh may be assigned a vertex ID for identifying each vertex.

  For example, when the input image is an image such as a JPEG or bitmap image and is expressed in a coordinate space defined by the position (x, y) and the luminance I, a mesh unit is used by using a conversion unit (not shown). To the original image. For example, an image such as JPEG or a bitmap image can be converted into the original image of the mesh using the method described in Non-Patent Document 1. Here, in addition to the xy plane as shown in FIG. 16, it may be a mesh represented by three-dimensional coordinates (x, y, I) with luminance I as the height.

  The mesh may be a two-dimensional mesh used in computer graphics (CG) texture mapping, and each vertex may have position information corresponding to the texture to be mapped, bitmap image, or the like. Good.

  When the input image is a color image, each vertex may further include color component information (for example, RGB).

  FIG. 3 is a block diagram illustrating the image processing apparatus 1. The image processing apparatus 1 includes an input unit 11, a detection unit 12, a setting unit 13, a calculation unit 14, a deformation unit 15, and an output unit 16. The deformation unit 15 includes a sorting unit 151 and a region deformation unit 152.

  The input unit 11 inputs an original image to be processed.

  The detection unit 12 detects one or a plurality of objects in the original image, and sets one target object region 301 from among the objects.

  When there are a plurality of object areas, the detection unit 12 may set the target object area by designation from the user. Alternatively, a known object recognition technique may be used to estimate whether the detected object area is a person or an object, and the person object area may be set as the target object area 301. Alternatively, the object area having the largest size may be set as the target object area 301.

The setting unit 13 sets a deformation area 302 that includes the target object area 301 and does not include another object area (the object area 304 in FIG. 1). In the present embodiment, the deformation area 302 is set based on designation from the user. For example, when the setting unit 13 receives an area specified by the user in the original image using a touch panel, a mouse, a keyboard, or the like (not shown), the area includes the target object area 301 and other object areas. If the determination is true, the designated area is set as the deformation area 302. If the determination is false, the user may be prompted to specify an area again. The object area determined not to be included in the deformation area 302 is not limited to a single object but may be an area with a high degree of attraction. 1, the object 303) may be included in the deformation area 302.

  The calculation unit 14 enlarges the target object area 301 in the deformation area 302 with an arbitrary position in the deformation area 302 as the center. In this embodiment, the deformation area 302 is distorted while enlarging the target object area 301 by moving the mesh vertex.

  In this case, an enlargement position (optimum position) at which the size of the enlarged target object region 301 is maximized and its enlargement magnification (maximum magnification) are calculated.

  That is, the calculation unit 14 selects a plurality of candidates for the enlargement position of the target object region 301 within the deformation region 302. Then, for each candidate, a maximum magnification that maximizes the enlargement magnification when the target object region 301 is enlarged so as not to protrude from the deformation region 302 around the enlargement position of each candidate is obtained. Among the plurality of maximum magnifications, for example, an enlargement position having the maximum magnification is selected as the optimum position.

  Note that the position of the object area in the present embodiment may be, for example, the position of the center of gravity of the object. In this case, the calculation unit 14 may calculate the centroid position from the shape of the object area.

  The sorting unit 151 sorts the deformation area 302 based on the moving direction of the target object area 301 to the optimum position. Each of the divided deformation regions 302 is subjected to different deformation processing by a deformation unit 152 described later.

  The area deforming unit 152 performs different deformation processing on each of the divided deformation areas 302 to move the target object area 301 to the optimum position, and the moved target object area 301 is centered on the optimum position. Magnify at maximum magnification. Thereby, the original image is deformed.

  The output unit 16 outputs the deformed original image (deformed image) to the presentation device 5. Here, the presentation device 5 includes a display, a printer, and the like. The presentation device 5 presents the deformed image as a thumbnail image.

  The input unit 11, the detection unit 12, the setting unit 13, the calculation unit 14, the deformation unit 15, and the output unit 16 may be realized by a central processing unit (CPU) and a memory used by the CPU.

  The configuration of the image processing apparatus 1 has been described above.

  FIG. 4 is a flowchart showing processing of the image processing apparatus 1.

  The input unit 11 inputs an original image to be processed (S101). The detection unit 12 detects one or a plurality of objects in the original image, and sets one target object region 301 from the detected one (S102).

  The setting unit 13 sets the deformation area 302 (S103). The calculation unit 14 calculates the optimum position and the maximum magnification when the target object area 301 is enlarged in the deformation area 302 around an arbitrary position in the deformation area 302 (S104).

  The sorting unit 151 sorts the deformation area 302 based on the moving direction when moving the target object area 301 from the current position to the optimum position (S105).

  The deformation unit 152 deforms the original image (S106). That is, the deforming unit 152 performs different deformation processing on each of the divided deformation areas 302 to move the target object area 301 to the optimum position, and the moved target object area 301 around the optimum position. Is enlarged at the maximum magnification (S106). The output unit 16 outputs the deformed original image (deformed image) to the presentation device 5 (S107).

  The processing of the image processing apparatus 1 has been described above.

  Hereinafter, each part is explained in full detail.

  The input unit 11 supplies the original image to the detection unit 12, the setting unit 13, and the deformation unit 15. The detection unit 12 detects the target object region 301 in the original image. For example, the detection unit 12 may detect a set of patches including an area extracted by the method of Patent Literature 2 as the target object area 301.

  The detection unit 12 notifies the target object region 301 to the setting unit 13 and the calculation unit 14. For example, the detection unit 12 may notify the setting unit 13 and the calculation unit 14 of the patch ID of the patch included in the target object region 301.

  The setting unit 13 sets the deformation area 302. The setting unit 13 notifies the calculation unit 14 of the deformation area 502. For example, the setting unit 13 may notify the calculation unit 14 of the patch ID of the patch included in the deformation area 502.

  The calculation unit 14 calculates the optimum position and the maximum magnification when the target object area 301 is enlarged in the deformation area 302 around an arbitrary position in the deformation area 302.

  FIG. 5 is a flowchart showing the processing of the calculation unit 14. Here, the position of the target object area 301, the enlargement magnification M, the movement direction (angle A), and the movement distance L are increased by a certain step size, and the target object area is within a range that falls within the deformation area. The optimum position P, the maximum magnification M, the moving direction A, and the moving distance L that can enlarge 301 to the maximum are calculated.

  In step S201, the calculation unit 14 initializes a magnification M for enlarging the target object region 301 (S201). For example, it may be initialized to M = 1.0.

  In step S202, the calculation unit 14 initializes the direction (angle A) in which the target object region 301 is moved (S202). For example, A ′ = 0 ° may be initialized. Here, the angle A ′ is an angle from the reference line in the original image. FIG. 6 is an explanatory diagram of the processing of the calculation unit 14. As shown in FIG. 6A, for example, the reference line may be a line parallel to the horizontal direction of the original image, and the counterclockwise direction may be positive and the clockwise direction may be negative. In this case, the direction of the reference line may be an angle of 0 °.

  In step S203, the calculation unit 14 determines whether or not the angle A is less than 360 ° (S203). If the angle A ′ is less than 360 ° (determination is YES), the process proceeds to step S204. If the angle A ′ is 360 ° or more (determination is NO), the process proceeds to step S205.

  In step S204, the calculation unit 14 initializes a distance L for moving the target object region 301 (S204). For example, it may be initialized to L ′ = 0.

  In step S206, the calculation unit 14 adds a predetermined step distance ΔL to the distance L ′ as shown in FIG. 6B (S206).

  In step S207, the calculation unit 14 moves the target object region 301 by a distance L ′ in the direction of the angle A ′ (S207). That is, the vertex of the mesh included in the target object region 301 is moved by the distance L ′ in the direction of the angle A ′.

  In step S208, it is determined whether or not the moved target object area 301 is included in the deformation area (S208).

  In the present embodiment, if the target object area 301 and the deformation area 302 do not intersect and one point constituting the boundary of the target object area 301 is inside the deformation area 302, the target object area 301 is in the deformation area 302. It is determined that it is included.

  Specifically, by searching for the target object area 301 that is adjacent to a patch that is not registered as the target object area 301 among the vertices of the set of patches registered in the memory (not shown). The mesh vertices U11, U12,..., U1n constituting the boundary of the target object region 301 can be obtained (the same applies to the deformation region 302).

  Further, ridgelines T11, T12,..., T1i... Constituting the boundary of the target object area 301 can also be obtained (T21, T22,... T2j are similarly applied to the deformation area 302). Here, if there is no combination that intersects the ridgelines T1i and T2j of the boundary between the target object area 301 and the deformation area 302, it is determined that the target object area 301 and the deformation area 302 do not intersect. Furthermore, when the ridge line of the deformation area 302 does not intersect with the line segment drawn to U11 from a certain point inside the patch belonging to the deformation area 302 or even times, the target object area 301 becomes the deformation area 302. It is determined that it is included.

  If the moved target object area 301 is included in the deformation area, the process proceeds to step S210. If not included, the process proceeds to step S209.

  In step S209, the calculation unit adds a predetermined step angle ΔA to the angle A ′ (S209), and proceeds to step S206.

  In step S210, when the calculation unit 14 can enlarge the target object region 301 to the maximum at a position where the target object region 301 is moved by a distance L ′ in the direction of the angle A ′ from the initial position in the input original image. The enlargement center P ′ and the magnification M ′ at that time are calculated.

  FIG. 7 is a flowchart showing the process of the calculation unit 14 in step S210. The calculation unit 14 sets a plurality of candidate points P′m (m = 1 to s) (s is the number of candidate points) that are candidates for the expansion center P ′ in the target object region 301 (S301). .

  FIG. 8 is an example diagram showing candidate points P′m. As shown in FIG. 8A, for example, the calculation unit 14 sets candidate points P′m in a lattice pattern at predetermined intervals from the center of gravity of the target object region 301 in the target object region 301. Also good.

  The calculation unit 14 sets m = 1 (S302). The calculation unit 14 enlarges the target object region 301 with the enlargement factor M′m with the candidate point P′m as the enlargement center (S303).

  Specifically, as shown in FIG. 8B, a point V ′ (P) where a half line P′mV connecting the expansion center P′m to the vertex V of the mesh in the target object region 301 is M times larger. The vertex V of the mesh is moved to “mV” = M′m × P′mV). By performing this operation on all mesh vertices in the target object region 301, the target object region 301 is enlarged.

  The calculation unit 14 determines whether or not the enlarged target object area 301 is included in the deformation area 302 (S304). When the determination in step S304 is YES, the calculation unit 14 newly sets a value (M′m + ΔM ′) obtained by adding a predetermined constant ΔM ′ (for example, ΔM ′ = 0.1) to the enlargement magnification M′m. As the enlargement magnification M′m (S305), the process proceeds to step S303.

  When the determination in step S304 is NO, the calculation unit 14 determines the enlargement magnification M'm (S306). For example, the calculation unit 14 may determine M′m = M′m−ΔM as the final enlargement magnification at the candidate point P′m.

  The calculation unit 14 determines whether or not the current enlargement magnification M′m is larger than the enlargement magnification M′m before the update (S307). When the determination in step S307 is NO, the calculation unit 14 determines whether m = s (S311). That is, the calculation unit 14 determines whether or not the evaluation is performed for all P′m. When the determination in step S311 is NO, (m + 1) is set as a new m, and the process proceeds to step S303.

  When the determination in step S309 is YES, the calculation unit 14 writes the enlargement magnification M′m as a new enlargement magnification M ′ in a memory (not shown), and updates the enlargement magnification M. Further, P′m at that time is written as a magnification center P ′ in a memory (not shown), and the magnification center P ′ is updated (S310). And it changes to step S311.

  If the determination in step S311 is YES, the process proceeds to step S211 in FIG.

  In step S211, the calculation unit 14 determines whether or not the calculated enlargement magnification M ′ is larger than the enlargement magnification M that has been updated so far (S211). If the determination is YES, the calculation unit 14 determines in step S212. The memory (not shown) is updated with the enlargement magnification M ′ as the enlargement magnification M. That is, the calculation unit 14 calculates the enlargement magnification M ′ when the movement distance L ′ is the angle A ′, and updates the memory (not shown) with the enlargement magnification M ′ having the maximum value as the enlargement magnification M. To do. At this time, the memory (not shown) is updated with the angle A ′ as the angle A and the distance L ′ as the distance L. When the determination in step S211 is NO, the process proceeds to step S206.

  In step S205, the calculation unit 14 sets the optimum position P, the maximum magnification M, the movement direction A, and the movement distance L, and ends the process.

Although the method for automatically calculating the optimum position, maximum magnification, moving direction, and moving distance has been described here, the user may give these values from the outside within the range where the target object is within the deformation affected area. . In this case, the angle A ′, the distance L ′, the position P ′, and the magnification M ′ are fixed as given from the outside, and the process of FIG. 5 is performed.

  The processing of the calculation unit 14 has been described above.

  The sorting unit 151 moves the target object region 301 from the current position to the optimal position P based on the optimal position P, the maximum magnification M, the moving direction A, and the moving distance L. Based on the trajectory, the deformation region 302 is segmented.

  FIG. 9 is an explanatory diagram of the processing of the sorting unit 151. Based on the target object area 301, the deformation area 302, and the movement direction A, the sorting unit 151 is parallel to the movement direction A and touches the object area 301, and intersects with the outer circumference of the deformation area 302. pull. That is, the sorting unit 151 divides the deformation region 302 into a region 305, a region 306, a region 307, and a region 308 by the straight line 303 and the straight line 304. The sorting unit 151 gives an attribute (region ID) that can determine which region of the regions 305 to 308 belongs to the vertexes included in each region.

  Note that the straight line 303 and the straight line 304 pass through any one of a plurality of points constituting the boundary of the target object region 301, and are most distant from each other among a plurality of straight lines parallel to the moving direction A of the target object region 301. It is calculated as two separate straight lines.

  Further, the sorting unit 151 may obtain the area ID as follows. The sorting unit 151 assigns the patch to the (i) region 308, (i) based on the positional relationship between the straight line parallel to the moving direction A of the target object region 301, the straight line 303, and the straight line 304. ii) Classification is made into three categories: area 305 or 307, and (iii) area 306.

  If (ii) it belongs to the region 305 or the region 307, the sorting unit 151 determines that the patch is the region 305 or the region from the position of the intersection between the patch and the ridge line that forms the boundary of the target object region 301. Which area 307 belongs to is determined. For a patch existing on the straight line 303 or the straight line 304 that is the boundary of the divided areas, the area ID of any one area in contact with the patch is given.

  In addition, after drawing the straight line 303 and the straight line 304, the method of dividing | segmenting a mesh so that the line segment may be considered is also considered. Specifically, a patch (a triangle) through which the straight line 303 and the straight line 304 pass is divided on the line through which the line segment passes. When a quadrangular patch is generated by the division, the quadrangular patch is divided into two triangular patches by a method such as Delaunay triangulation described in Non-Patent Document 1.

  The processing of the sorting unit 151 has been described above.

  The deformation unit 152 performs different deformation processing on each of the divided deformation areas 302 to move the target object area 301 to the optimal position P, and the target object area 301 moved around the optimal position P is the center. Is enlarged at the maximum magnification M.

  FIG. 10 is a flowchart showing the process of the area deforming unit 152. As shown in FIG. 10, in step S <b> 401, the sorting unit 151 determines which region of the regions 305 to 308 each vertex belongs to based on the region ID assigned by the sorting unit 151. (S401). In step S402, the region deformation unit 152 moves the target object region 301 by moving the vertex for each region (S402). In step S403, the area deforming unit 152 enlarges the target object area 301 (S403). Hereinafter, steps S402 and S403 will be described.

  Step S402 will be described.

  The area deforming unit 152 moves the target object area 301 to the optimal position P. That is, the target object area 301 is moved in the movement direction A by the movement distance L. At this time, the region deforming unit 152 performs interpolation by expanding the region 305 in the moving direction A and contracting the region 307 in the moving direction A as the target object region 301 moves. Further, the area deforming unit 152 performs interpolation by shifting the area 306 and the area 308 in the movement direction A as the target object area 301 moves.

  Hereinafter, a method for moving mesh vertices for realizing the above-described deformation will be described.

The mesh vertices in the target object region 301 are realized by moving all the mesh vertices in the movement direction A by the movement distance L.

  FIG. 11 is an explanatory diagram of the deformation process for the area 305. The region deforming unit 152 draws a straight line F parallel to the moving direction A of the target object region 301 from the mesh vertex W to be moved. Here, the intersection of the outer periphery of the target object region 301 and the straight line F is R1, and the intersection of the outer periphery of the deformation region 302 and the straight line F is R2 (FIG. 11A). Further, the intersection of the outer periphery of the target object area 301 after movement and the straight line F is defined as R1 '(FIG. 11 (b)).

  At this time, as R1 moves to R1 ', the vertex W moves so as to linearly interpolate between R1 and R2. That is, the region deforming unit 152 moves the vertex W to a point W ′ where the line segment R2W: line segment WR1 = line segment R2W ′: line segment W′R1 ′. The area deforming unit 152 deforms the original image by performing the same process on all the mesh vertices in the area 305. The region deforming unit 152 performs the same process on the mesh vertices in the region 307.

  FIG. 12 is an explanatory diagram of the deformation process of the area 306. The point of contact between the straight line 303 dividing the deformation area 302 and the target object area 301 is Q1, and the intersection of the straight line F 'and the outer periphery of the deformation area 302 is Q2. The straight line F ′ is a straight line passing through the mesh vertex U to be moved and the contact point Q1. Further, a point corresponding to the point Q1 of the target object area 301 before the movement on the target object area 301 after the movement is defined as Q1 '.

  The region deforming unit 152 moves the point Q1 to the point Q1 'and moves the point U so as to linearly interpolate between Q1 and Q2. That is, the point U is moved to the point U ′ where Q2U: UQ1 = Q2U ′: U′Q1 ′. The area deforming unit 152 deforms the original image by performing the same process on all the mesh vertices in the area 306. The area deforming unit 152 performs the same process for the mesh vertices in the area 308.

  When the target object region 301 is in contact with the straight line 303 at a plurality of points or sides, the region deforming unit 152 may set a point on the contact or tangent closest to the vertex U to be moved as Q1.

  Step S403 will be described.

  The area deforming unit 152 enlarges the moved target object area 301 at the magnification M with the optimum position P as the enlargement center. That is, the area deforming unit 152 enlarges the target object area 301 by moving the mesh vertices of the image after the movement of the target object area 301 according to the enlargement magnification M. FIG. 13 is a flowchart showing enlargement processing of the target object area 301 by the area deforming unit 152. FIG. 14 is an explanatory diagram of the enlargement process of the target object area 301.

  In step S501, the region deforming unit 152 selects one vertex that has not been selected in the moved target object region 301 as a processing target vertex (processing target vertex V) (S501).

  In step S <b> 502, the region deforming unit 152 draws a half line PV from the optimum position P to the processing target vertex V, and an intersection D <b> 1 between the half line PV and the outer periphery of the target object region 301, and the half line PV and the deformed region 302. An intersection point D2 with the contour is obtained (S502).

  The intersection D1 between the half line PV and the target object area 301 and the intersection D2 between the half line PV and the deformation area 302 have the relationship shown in FIG.

  In step S503, the region deformation unit 152 determines whether or not the selected movement target vertex V is in the target object region 301 (S503). When the determination in step S503 is YES, in step S504, the region deformation unit 152 selects the selected moving target vertex at the position of the point V ′ where “PV ′ (vector) = enlargement magnification M × PV (vector)”. V is moved (S504).

  In step S507, the region deforming unit 152 determines whether all mesh vertices have been selected in the image after movement of the target object region 301 (S507). For example, the region deforming unit 152 may perform the determination in step S507 by storing the vertex ID of the vertex selected as the movement target vertex V in a memory (not shown).

  When the determination in step S507 is YES, the region deformation unit 152 notifies the presentation unit 16 that the image deformation has ended, and the process ends. If the determination in step S507 is no, the process proceeds to step S501.

  When the determination in step S503 is NO, in step S505, the region deformation unit 152 determines whether or not the selected movement target vertex W is inside the deformation region 302 and outside the target object region 301 (S505). If the determination in step S505 is no, the process transitions to step S507. When the determination in step S505 is YES, in step S506, the moving target vertex V is moved to the position of the point V ′ where “D1′V ′: V′D2 = D1V: VD2” (S506). Here, D <b> 1 ′ is an intersection of the outer periphery of the target object region 301 enlarged at the enlargement magnification M and the half line PV ′. And it changes to step S507. Thereby, a deformed image for making a thumbnail image is generated (FIG. 14B).

  The output unit 16 outputs the deformed image to the presentation device 5.

  The presentation device 5 reduces the deformed image and presents it as a thumbnail image. For example, the presentation unit 16 may be a display that renders and displays thumbnail images. In this case, the presentation device 5 may perform rendering by interpolating the inside of the patch using color information of each vertex of the mesh (see Non-Patent Document 1).

(Modification)
In the present embodiment, the original image is represented by a mesh, but may be a pixel image. In this case, the detection part 12, the setting part 13, and the calculation part 14 should just perform the same process by making the vertex of the above-mentioned mesh into the center point of each pixel.

  The deforming unit 15 performs image deformation by rewriting the luminance and color information of each pixel stored in a memory (not shown). That is, the movement destination of the center point of each pixel is temporarily calculated in the same manner as the method for determining the movement destination of the vertex of the mesh described above. The coordinates calculated as the destination of the center point of each pixel are given the pixel brightness and color information as a reference point, and the reference point is interpolated to rewrite the brightness and color information of each pixel. Generate an image of

  According to the present embodiment, it is possible to generate a thumbnail image that allows the user to easily specify the original image.

(Second Embodiment)
The image processing apparatus according to the second embodiment is different from the previous embodiment in that the setting unit 13 automatically sets the deformation area 302.

  In the present embodiment, the setting unit 13 sets the deformation area 302 so as to include the target object area 301 and not include other objects (object 304 in FIG. 1). For example, the deformation region 302 may be specified using a known active contour model (Snake method) described in Non-Patent Document 2.

  FIG. 15 is a flowchart showing processing of the setting unit 13 in the present embodiment.

  In step S601, the setting unit 13 prepares control points C1 to Cn at positions of mesh vertices on the outer periphery of the target object 301, and sets a contour obtained by connecting them as a contour C (S601).

  In step S602, the setting unit 13 sets an initial magnification r (S602).

  In step S603, the setting unit 13 enlarges (moves) the control points C1 to Cn of the contour C with the magnification r about the center of gravity of the contour C (S603).

  In step S604, the setting unit 13 determines whether there is a control point that protrudes from the mesh (S604). When the determination in step S604 is YES, in step S605, the setting unit 13 reduces the magnification r by a predetermined step size (S605).

  When the determination in step S604 is NO, in step S606, the setting unit 13 moves the control point to a position where the energy (E = αEint + βEimg) is decreased. Here, Eint is a shape energy term (internal energy term), Eimg is an image energy term (external energy term), and α and β are weights for each energy term.

  That is, the setting unit 13 moves the control point so that the contour C is short, smooth, and is positioned along a ridge line where the dihedral angle of the adjacent patch is small.

  In step S607, the setting unit 13 determines whether all the control points have been moved (S607). If the determination in step S607 is no, the process transitions to step S606.

  When the determination in step S607 is YES, in step S608, the setting unit 13 determines whether or not the processes in steps S606 and S607 have been repeated a predetermined number of times (S608). That is, it is determined whether or not the cycle of moving all of the control points C1 to Cn is repeated a predetermined number of times so that the energy is reduced. If the determination in step S608 is no, the process transitions to step S606.

  When the determination in step S608 is YES, in step S609, the setting unit 13 sets the patch area created by the mesh vertices included in the area surrounded by the contour C as the deformation area 302 (S609), and ends the process. To do.

  As described above, in this embodiment, since the deformation area 302 can be set so as to include the target object area 301 and not include other objects (the object 304 in FIG. 1), the user can specify the deformation area 302. Can be omitted.

  As described above, according to the above-described embodiments, it is possible to provide an image processing apparatus, method, and program that can generate a thumbnail image that allows a user to easily specify an original image.

  Although several embodiments of the present invention have been described so far, these embodiments are presented as examples and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

DESCRIPTION OF SYMBOLS 1 ... Image processing apparatus 5 ... Presentation apparatus 11 ... Input part 12 ... Detection part 13 ... Setting part 14 ... Calculation part 15 ... Deformation part 16 ... Output part 151 ... Classification part 152 ... Area deformation part

Claims (7)

A detection unit for detecting the first object region and the second object region from the original image to be processed;
A setting unit that sets, from the original image, a deformation area that includes the first object area and does not include the second object area;
Within the deformation area, a search unit that searches for an enlargement position where the size of the first object area can be enlarged, and calculates an enlargement magnification at the enlargement position;
The first object area is enlarged at the searched enlargement position at the enlargement magnification, and the remaining portion in the deformation area is subjected to deformation processing, and the second object area is not subjected to deformation processing or is deformed. An image processing apparatus comprising: a deformation unit that generates a deformed original image by reducing processing. The calculation unit includes:
The image processing apparatus according to claim 1, wherein a position where the size of the enlarged first object area is maximum is calculated as the enlarged position. The deformation part is
A partition unit that generates a partition region that partitions the deformation region based on a movement direction from the position of the first object region to the enlarged position;
Move in the moving direction, shrink the segmented area on the moving direction side, stretch the segmented area on the opposite side to the moving direction, and adjust the other segmented areas to the movement of the first object area Transform each said segmented area to transition,
The image processing apparatus according to claim 1, further comprising: a region deformation unit that expands the first object region at the expansion position while distorting the deformation region. The original image is
An image represented by a mesh that is a set of patches that are geometric planes in a coordinate space defined by position and brightness,
The region deformation portion is
The image processing apparatus according to claim 3, wherein each of the divided areas is deformed by moving each vertex of the patch in the divided area. The original image is
A pixel image represented by position and brightness;
The region deformation portion is
The image processing apparatus according to claim 3, wherein each of the partitioned areas is deformed by moving a vertex of a pixel in the partitioned area. A first object area and a second object area are detected from an original image to be processed;
From the original image, a deformation area that includes the first object area and does not include the second object area is set.
Within the deformation area, search for an enlargement position where the size of the first object area can be enlarged, and calculate an enlargement magnification at the enlargement position,
The first object area is enlarged at the searched enlargement position at the enlargement magnification, and the remaining portion in the deformation area is subjected to deformation processing, and the second object area is not subjected to deformation processing or is deformed. An image processing method for generating a deformed original image by reducing processing. Computer
Means for detecting a first object area and a second object area from an original image to be processed;
Means for setting, from the original image, a deformation area that includes the first object area and does not include the second object area;
In the deformation area, searching for an enlargement position where the size of the first object area can be enlarged, and calculating an enlargement magnification at the enlargement position;
The first object area is enlarged at the searched enlargement position at the enlargement magnification, and the remaining portion in the deformation area is subjected to deformation processing, and the second object area is not subjected to deformation processing or is deformed. An image processing program that functions as means for generating a deformed original image by reducing processing.

Images