WO2006075394A1 - 動きベクトル演算方法とこの方法を用いた手ぶれ補正装置、撮像装置、並びに動画生成装置 - Google Patents
動きベクトル演算方法とこの方法を用いた手ぶれ補正装置、撮像装置、並びに動画生成装置 Download PDFInfo
- Publication number
- WO2006075394A1 WO2006075394A1 PCT/JP2005/000424 JP2005000424W WO2006075394A1 WO 2006075394 A1 WO2006075394 A1 WO 2006075394A1 JP 2005000424 W JP2005000424 W JP 2005000424W WO 2006075394 A1 WO2006075394 A1 WO 2006075394A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- motion vector
- image
- resolution
- data
- calculating
- Prior art date
Links
- 239000013598 vector Substances 0.000 title claims abstract description 407
- 238000004364 calculation method Methods 0.000 title claims abstract description 67
- 238000000034 method Methods 0.000 title claims description 86
- 238000012937 correction Methods 0.000 title claims description 22
- 238000003384 imaging method Methods 0.000 title claims description 17
- 238000012545 processing Methods 0.000 claims abstract description 38
- 238000009499 grossing Methods 0.000 claims abstract description 20
- 238000012935 Averaging Methods 0.000 claims abstract description 5
- 238000000605 extraction Methods 0.000 claims description 9
- 230000006641 stabilisation Effects 0.000 claims description 8
- 238000011105 stabilization Methods 0.000 claims description 8
- 230000015572 biosynthetic process Effects 0.000 claims description 4
- 239000000203 mixture Substances 0.000 claims description 4
- 239000003381 stabilizer Substances 0.000 claims description 4
- 238000003786 synthesis reaction Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 description 49
- 238000010586 diagram Methods 0.000 description 18
- 239000002131 composite material Substances 0.000 description 8
- 238000013500 data storage Methods 0.000 description 7
- 239000000284 extract Substances 0.000 description 5
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 230000002194 synthesizing effect Effects 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/60—Control of cameras or camera modules
- H04N23/68—Control of cameras or camera modules for stable pick-up of the scene, e.g. compensating for camera body vibrations
- H04N23/682—Vibration or motion blur correction
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/80—Camera processing pipelines; Components thereof
Definitions
- Motion curve calculation method uses the method, imaging device, and moving image generation device
- the present invention relates to a technique for calculating a motion vector between a plurality of images, and more particularly to a camera shake correction apparatus, an imaging apparatus, and a moving image generation apparatus using this technique.
- Patent Documents 4 to 6 treat those having high correlation between motion vectors as having high reliability.
- the feature point of the image is grasped and the motion vector for the feature point is calculated.
- a polygon is formed by connecting points, recognizing how the polygon is deformed due to the movement of feature points between images to be compared, and estimating the movement position of other points. (For example, see Patent Documents 2 and 3).
- Patent Document 1 Japanese Patent Laid-Open No. 2003-78807
- Patent Document 2 Japanese Patent Laid-Open No. 11-506576
- Patent Document 3 Japanese Patent Laid-Open No. 10-341445
- Patent Document 4 Japanese Patent Laid-Open No. 9-73540
- Patent Document 5 JP-A-6-311502
- Patent Document 6 JP-A-6-153146 Disclosure of the invention
- the present invention can calculate a motion vector with high accuracy by a relatively simple process so that a computer with a small amount of data is not overloaded.
- An object of the present invention is to provide a motion vector calculation method capable of creating a high-quality photograph by correcting camera shake at the time of taking a picture, and a camera shake correction device, an imaging device, and a video generation device using this method. .
- a motion vector calculation method is a method for calculating a motion vector between images by inputting a plurality of image data. Processing for calculating and storing multiresolution data up to a predetermined predetermined resolution level L (L; natural number) for the image data, processing for calculating an input motion vector corresponding to the resolution level of the reference image, and A resolution-specific motion vector estimation process in which the following steps are executed using the reference image of the resolution L obtained by each process, the comparison image, and the input motion vector.
- L predetermined resolution level
- an energy function is defined with the coordinates on the reference image and the coordinates on the comparison image as arguments, and the energy is within a predetermined range of the input motion vector.
- the motion vector of resolution L is Calculating step, By using the calculated motion vector of resolution L as the input motion vector of higher resolution L 1 and repeating the above-described motion vector estimation process for each resolution for the higher resolution and image data in order. It is characterized by calculating the motion vector between the image and the comparison image.
- the degree of reliability means how reliable the motion vector is, or the degree to which the motion vector meets a predetermined condition, and is detected as an image feature amount by, for example, the color space of the image. It is determined by the value or function based on the image, such as the size of the edge (boundary) or the corner angle. Since these pieces of information represent image features, the reliability of motion vectors on the feature points (or lines and regions) is increased.
- the image feature amount includes a function using energy at the matching processing stage (for example, the reciprocal number), a function such that the portion estimated to be occlusion is 0, and the other is 1.
- the function is not limited to an arithmetic expression, but also includes a function that derives a corresponding numerical value by referring to a table.
- the reliability can be expressed by any one of the image feature amounts described above or any combination of image feature amounts (multiplication, addition, etc.).
- the degree of correlation is the degree that two points are estimated to move in the same direction, and is determined, for example, by the difference from surrounding motion vectors.
- the degree of correlation is not limited to this, and as a difference between image feature amounts, a function based on a difference in colors on two reference images, a difference in image features using a Gabor filter, or the like, or a user manually It can be expressed by a function using a specified cut (contour), and the difference in the value of the temporary motion vector It can also be expressed by multiplying or adding one or more of these functions, including functions based on.
- the image feature quantity used for the reliability calculation and the image feature quantity used for the correlation calculation may be different.
- each element of the input image has a value of RG B for all coordinates (x, y).
- I (x, y) (R (x, y), G (x, y), B (x, y))
- Each element of an image of resolution level L has multiple color information for all coordinates, such as (La * b * color space, YUV color space, edge image, corner-enhanced image), etc. Have information.
- the resolution level 0 is the original image and the resolution level L is increased, the image becomes blurred.
- Each element of the resolution level L is represented by I (x, y).
- Each element of the motion vector is a starting point position (X, y) and a motion vector v (x, y) force from the starting point position.
- the reason for having start point position information is that it does not have motion vectors for all coordinates (X, y).
- the motion vector includes a “grid type motion vector” having a motion vector for the start point coordinates arranged in a grid and a “point cloud type motion vector” having a motion vector of only characteristic points. There are s .
- a motion vector of resolution level L is obtained.
- the motion information has each element: ((x, y), v (x, y)) for each 2′L pixel.
- “′” represents a power.
- each motion vector v (x, y) is represented by r (x, y), for example, between 0 and 1 Have a value.
- the degree of correlation between motion vectors v (xl, yl), v (x2, y2) is represented by a (xl, yl, x2, y2), for example, having a value between 0 and 1 Like that.
- Fig. 1 the data of the start position (x, y), the motion beta v (x, y), and the reliability r (x, y) corresponding to each grid (grid) of the reference image To have.
- data of correlation degree a (xl, yl, x2, y2) is provided in association with any two grid points (or a line segment determined by two points).
- the reliability of the motion vector of a grid (for example, edge or corner) having a large amount of features in the reference image is set to be high.
- the degree of correlation between two motion vectors with a large difference in color and a large difference in provisional motion vector is set to be low.
- the reliability of the motion vector at the corner or edge of the image is increased at the discontinuous location indicated by the dotted line A in FIG.
- the degree of correlation between motion vectors across discontinuous locations is low.
- circles indicate grids, and the thickness of the round frame indicates the level of reliability.
- B and E are the reliability associated with the grid. E is an edge, indicating that the reliability is high, and B is neither a corner nor an edge, indicating that the reliability is low.
- C and D are the correlations associated with the grid. C is a high correlation because the difference in motion vectors is small, and D is a low correlation because the difference in motion beta is large. Show me that.
- a matching step for temporarily calculating a motion vector having the reliability and the correlation as defined above as parameters, and a motion vector within a predetermined range of the temporarily determined motion vector are considered. Then, the motion vector is calculated by a smoothing process that is adjusted by calculating the affine parameter by weighted averaging or least square method, and this is executed on the high resolution image in order from the low resolution image. Since it can be executed by so-called repetitive calculations, it is possible to calculate a motion vector with high accuracy without imposing a load on the computer.
- the motion vector calculation method includes a process of inputting instruction information related to the feature point of the reference image, and the motion between the reference image and the comparison image based on the input instruction information. It is characterized by calculating a vector.
- the motion vector is calculated based on the feature point instructed by the user, and this motion vector is taken as a non-moving one, and the other motion vector is calculated.
- the vector can be calculated easily.
- the motion vector calculation method further extracts a plurality of feature points of the reference image in a distributed manner, and based on the extracted feature points, the motion vectors between the reference image and the comparison image. It is characterized by calculating the spectrum.
- distributed means that when an image feature point is generated at a certain position, another image feature point is not generated in the vicinity of the image feature point.
- the motion vector further includes a corresponding feature point between the reference image and the comparison image, a line connecting the feature point, and the feature point or It is motion vector mesh data representing each motion of a region constituted by the lines.
- the region of the motion beta mesh data may be divided into one or more polygons, particularly triangles.
- the motion between lines is managed by representing the motion of the line segment and the region only by the association of the points with motion vector mesh data. As a result, it is possible to manage movement with high precision and accuracy with a small amount of data.
- a camera shake correction apparatus is a camera shake correction apparatus that performs camera shake correction using the motion vector calculation method described above, and inputs a plurality of images to be one or more than the reference image.
- the motion vector with the comparison image is calculated using the motion vector calculation method, and the motion vector calculation unit that outputs the calculation result is compared with the motion vector obtained by the motion vector calculation unit.
- a composite image generating means for creating a composite image by executing a process for superimposing the reference image on the reference image, and one or two or more composite images created by the composite image generating means Combine images on Image synthesizing means, and image data output means for outputting image data synthesized by the image synthesizing means.
- the image composition unit calculates a maximum range of camera shake, and performs an averaging process based on the number of images for the maximum range from the image frame.
- an imaging apparatus is an imaging apparatus having the above-described camera shake correction device, and continuously captures an image of a subject, generates a plurality of images, and inputs the images to the camera shake device. And a storage unit for storing an image of the camera shake device.
- a moving image generation apparatus is a moving image generation apparatus that generates a moving image using the motion vector calculation method described above, and receives a plurality of images and uses the motion vector calculation method as a reference.
- a motion vector calculation unit that calculates a motion vector between an image and one or more comparison images, and a motion vector calculated by the motion vector calculation unit is divided by an arbitrary predetermined value or
- An intermediate image is generated by applying the intermediate motion vector to a reference image and a motion vector editing unit that calculates two or more intermediate motion vectors, and an initial image, an intermediate image, and a final image are And a moving image generating means for enabling sequential display.
- the program according to the present invention is a program for inputting a plurality of image data and calculating a motion vector between the images.
- a predetermined resolution level L A process of calculating and storing multi-resolution data up to (L; natural number), a process of calculating an input motion vector of resolution level L of the reference image, and a reference image of resolution L obtained by each of the above processes, comparing (1) For each start point coordinate of the input motion vector, the coordinates on the reference image and the comparison image are displayed.
- a matching step for temporarily calculating a motion vector using reliability and correlation as parameters, and smoothing for adjusting the temporarily determined motion vector according to the state of the motion vector within a predetermined range.
- the motion vector can be calculated by performing a so-called iterative operation in which the motion vector is calculated and executed on the high-resolution image in order from the low-resolution image. It is possible to calculate high motion beta.
- FIG. 2 is a block diagram of a motion vector calculation apparatus according to the present embodiment. This apparatus inputs a plurality of images and calculates a motion vector between the images.
- the motion vector calculation device 1 includes an external storage device 10 such as a CD-ROM drive for reading and writing image data, a keyboard, a mouse, etc. for inputting user (operator) instruction information.
- a display unit 12 that displays image data
- a motion vector calculation unit 13 that calculates a motion vector based on the input image data
- a storage unit that stores the input image data and motion vector data Has 15.
- the motion vector calculation unit 13 receives image data input means (function) 31 that inputs image data from the input device 11, the display device 12, and the external storage device 10 and stores the image data in the storage unit 15.
- Multi-resolution image generation means (function) 32 that filters each image data to create image data at multiple resolution levels (function) 32;
- Mesh specification means that specifies a motion vector mesh based on instructions from the input device 11 (function) 33)
- Image feature point group extraction means (function) 34 that extracts one or more feature points of the image 34
- Triangulation means (function) 35 that divides the figure connecting the feature points of the image into triangles It has motion vector estimation means (function) 36 for calculating the motion vector based on the feature points, and motion vector output means (function) 37 for outputting the calculated motion vector data.
- Each of the means 31 and 37 is executed as a CPU function.
- the image data is data having w x h grids with width w and height h, and pixel values (eg, RGB, YCbCr) in each grid.
- the image feature point group data is a set of feature point data of the image data, and each image feature point is composed of its coordinate value (x, y).
- the motion vector mesh data is data of the motion beta of the image. It consists of multiple motion vector points, motion vector lines, and motion vector areas.
- motion vector mesh data can be set for the image in Fig. 3 (a) as shown in Fig. 3 (b).
- the motion vector point is composed of the coordinates (x, y) on the image, the starting point of the motion beta, the motion vector (dx, dy), and the determined flag fp.
- the determined flag is a flag indicating whether or not the motion vector at the point has been determined.
- An example of motion vector point data is shown in Fig. 4 (a). Coordinates, motion vectors, and determined flags are stored in association with each motion vector point identification information (ID).
- the motion vector line is composed of pointers [pl, p2] to two motion vector points which are the end points and a determined flag fe.
- the determined flag is a flag indicating whether or not the moving solid on the line has been determined.
- the motion vector region is a pointer [el, e2, en] to the motion vector line that is the boundary (in the case of 3 ⁇ 4 boundary force), or Vertex movement It consists of pointers to vector points [pl, p2, ..., ⁇ ] (when the number of vertices is ⁇ ) or both. It also has a determined flag fr indicating whether the motion vector in the area has been determined or not defined. This area is a triangle, but any polygon can be used.
- the image data input from the external storage device 10 is stored in the input image data file 50 of the storage unit 15 by the image data input means 31.
- the input image is filtered by the multi-resolution generation means 32 to create a low-resolution image and stored in the multi-resolution image data file 51.
- FIG. 5 is a flowchart showing the processing procedure of the multi-resolution image generating means 32.
- the input image is subjected to color space conversion, edge image generation processing, etc. (S101) to create a resolution level 0 image, and each of the resolution level 0 images is subjected to a blur filter and an average value filter.
- Etc. are applied to create low-resolution image data (S102), and the created multi-resolution image data is stored in the multi-resolution image data file 51.
- FIG. 6 is an example of multi-resolution image data.
- vertical line A represents an image converted to Lab color space
- line B represents an edge image only in the vertical direction
- line C represents an edge image only in the horizontal direction
- line D represents an edge image having no direction dependency.
- the horizontal direction represents the resolution level to which the force filter is applied. The higher the resolution level value, the lower the resolution.
- the user activates the mesh designating means 33, extracts the reference image and the comparison image, which are the reference for motion vector calculation, from the input image data file 50 using the input device 11 and the display device 12, and outputs both images.
- the corresponding part of is specified as a point, line, or area.
- a plurality of comparison images may exist. Or you can set the grid only with points. Les.
- the mesh designating means 33 creates point, line, and area motion vector mesh data in the format shown in FIG. 4 by inputting this designation information, and stores it in the motion vector mesh data A file 52. At this time, in the motion vector mesh data A, the determined flags ft, fe, fr of all motion vector point “line” regions are set.
- the image feature point group extraction means 34 is activated through the input device 11 or upon completion of the operation of the mesh specifying means 33, and inputs the multi-resolution image data and the motion vector mesh data A to obtain the image feature point group data. Is output.
- the image feature points are generated around the edges and corners of the image. When an image feature point is generated at a certain position, another image feature point is not generated in the vicinity. This is because the motion vector at a certain position in the image is expected to have almost the same value as the motion vector in the vicinity. In this way, image feature points are distributed so that the overall motion vector can be approximated with as few points as possible.
- the motion vector points in the motion vector mesh data A 'line are not generated on or near the region. This is because there is no need for automatic calculation because the user explicitly specifies the motion vector.
- the image feature points for the motion vector mesh data A in FIG. 3 (b) or FIG. 7 (b) are as circles in FIG. Even if feature points are detected on and around the motion vector mesh, they should not be saved as feature points.
- the triangulation unit 35 receives the image feature point group data 53 and the motion vector mesh data A as inputs, and triangulates the region based on the feature points to obtain a motion vector message. Output data B and save it in motion vector mesh data B file 54.
- Triangulation is performed by adding the image feature point group while retaining each data of the motion vector point 'line / region possessed by the motion vector mesh data A.
- An example of such a triangulation is a constrained Delaunay triangulation, for example.
- the triangulation unit 35 sets the determined flag if the motion vector mesh point originally exists in the motion vector mesh data A, while the image feature In the case of motion vectors automatically added by the point cloud extraction means 34, do not set fp. Similarly, regarding the determined flag fe and fr of the motion vector line 'region, set it if it is included in the motion vector mesh data A, and set it if it is automatically generated by the triangulation means 35. Nare ,.
- the motion vector mesh data A in FIG. 8 and the motion vector mesh data B obtained as a result of triangulation for the image feature point group are as shown in FIG.
- the square and thick line parts of the mesh are the parts corresponding to the motion vector mesh data A and are specified by the user, so the determined flag is set, and the other round points and thin line parts are determined flags. Is not set.
- the motion vector estimation means 36 receives the multi-resolution image data 51 and the motion vector mesh data B as inputs, and automatically calculates the motion vectors for the motion vector points for which the determined flag fp is not set.
- the motion vector mesh data C with the determined flag of the point * line * area set is output and saved in the motion vector mesh data C file 55.
- the motion vector of the square point 'bold line' is already given by the user and the determined flag is set, so there is no need to calculate it. Only the part is subject to automatic calculation. In this automatic calculation, considering the consistency with the motion vector of the square / thick line part, the calculation is executed by the following processing method without changing the motion vector of the square point / thick line part.
- the motion vector saved in the motion vector mesh data B file 54 is set as an initial motion vector, and resolution correction processing is performed on this (S201).
- resolution correction processing motion vectors are thinned out according to the resolution. For example, in the case of a grid-type motion vector with a resolution of L, motion information is divided into elements ((x, y), v (x, y) have).
- a motion vector (including mesh data) that is a motion vector of a point not on the lattice and for which a determined flag is set is Are associated as motion vectors on neighboring grids according to a predetermined movement rule. At this time, if there are motion vectors moving from multiple points for the same grid, the average value of these motion vectors is associated as the motion vector of the grid.
- a motion vector for which the determined flag is not set is processed as (0, 0) in the case of, for example, two dimensions.
- motion vector superimposition processing is executed (S202).
- all the (0, 0) motion vector data is superimposed on the motion vector data of resolution L (initial resolution) created in step S201.
- resolution L initial The resolution (motion) motion vector data becomes the output of this step as it is.
- a matching process is executed for the motion vector of resolution L obtained by correction in this way (S203).
- FIG. 11 is an explanatory diagram of the matching processing procedure when there are two images.
- the comparison image, and the input motion vector using the reference image of the resolution L obtained by each of the above processes, the comparison image, and the input motion vector, first, for each start point coordinate of the input motion vector, the coordinates on the reference image and the comparison image An energy function having coordinates as arguments is defined, and a motion vector is temporarily calculated based on the energy function within a predetermined range of the input motion vector (S301). Further, for each feature point of the reference image, reliability is provisionally calculated using information such as edges and corners (S302). The reliability may be performed using the energy function described above in addition to a function based on an image such as the size of an edge. How to set the reliability depends on what conditions are important in the implementation. For example, in an application where matching is focused on the center of the screen, smoothing processing may be performed using the Gaussian function with the center of the screen as the reliability.
- the correlation is calculated based on the color difference between the feature points of the reference image and the difference between the temporarily calculated motion vectors (S303).
- the degree of correlation is, for example, multiplied by the square of the pixel color difference and the square of the magnitude of the motion vector difference, and is set to 1 to 10 using a Gaussian function or a similar function.
- the motion vector, reliability, and correlation degree data are calculated for each of the reference image and the plurality of comparison images as shown in FIG. Also good. In this case, there is a case where the movement of each comparison image with one reference image is calculated, and that each comparison image becomes a reference image between the next comparison image. [0079] Further, so-called block matching may be used as an example of the energy function.
- smoothing processing is executed after the matching processing in step S203, and a motion vector of resolution L is fully calculated (S204).
- FIG. 13 shows the procedure of the smoothing process.
- the motion map and reliability temporarily calculated by the matching process are averaged (smoothed) by the weighted average process using the nearby motion vector, reliability, and correlation. Calculate the L motion vector.
- the motion vector is a weighted average based on the product of the reliability and the correlation
- the reliability is a weighted average based on the correlation
- the affine parameters may be obtained by the least square method instead of the above weighted average processing.
- motion vector data with resolution L 11 is generated by linear interpolation using the motion vector with resolution L obtained as a result of the smoothing process (S207).
- v (32,36) 1/2 v (32,32) + 1/2 v (32,40)
- v (36,36) 1/4 v (32,32) + 1/4 v (32,40) + 1/4 v (40,32) + 1/4 v (40,40) And calculate.
- step S202 the interpolated motion vector data is superimposed on the input motion vector data created in step S201 to obtain an input motion vector for the matching process.
- the motion vector output means 37 outputs the motion vector data stored in the motion vector mesh data C file 55 to the external device 10 or other program module.
- a temporary motion vector is obtained using only color information, and a different concept of a measure based on the difference between the reliability based on the edge information and corner information and the correlation between the motion vectors. Therefore, it is possible to calculate a motion vector with high accuracy without imposing a load on the computer.
- calculations can be performed according to the computer's ability and the required accuracy of motion vectors, so image stabilization and video generation that are highly versatile. It can be used for various purposes.
- the motion vector computing device 1 described above can also be implemented by the configuration shown in FIG. That is, the image data input by the image data input means 31 is used to create multi-resolution image data by the multi-resolution image generation means 32, and using this, the motion vector estimation means 36 uses the above processing to produce a grid type motion vector. Is calculated.
- the functions of the mesh designating means 33, the image feature point group extracting means 34, and the triangulation means 35 in FIG. 2 can be activated depending on the load and application of the computer.
- the mesh designation means 33 can be omitted.
- the image feature point group extraction means 34 extracts feature points from the edge image of resolution level 0 as multi-resolution image data to create image feature store group data, and based on this, the triangulation means 35 Create a motion solid mesh data B. Then, the motion vector estimation means 36 calculates a motion vector using the motion vector mesh data B.
- the image feature point group extraction means 34 may execute feature point extraction processing using image data having a low resolution in the multi-resolution image data.
- image data having a low resolution When an extremely large number of feature points are extracted from the input image (resolution level 0), it is possible to extract only points with a large feature amount by extracting feature points at a low resolution. For this reason, it is possible to reduce the load on the computer compared to selecting and comparing feature quantities of many feature points of image data with high resolution.
- This motion vector arithmetic unit 1 inputs a plurality of images, outputs a motion vector with respect to a reference image, manages image data by motion vectors, and compresses a series of motion images.
- the apparatus of the following embodiment can be configured by using the motion vector calculation unit 13 and the main part of the motion vector data storage unit 15 of the apparatus 1.
- the motion vector computing device 1 it is also possible to provide a means for setting a search range by using the function of the mesh designating means 33. That is, the mesh designating means 33 designates corresponding points between the reference image and the comparison image, obtains a motion vector between them, and sets a determined flag. Only the motion vector data is set without setting this determined flag. Calculate and store. Then, the motion vector estimation means 36 searches for a region in a predetermined range of the stored motion beta, and calculates a motion vector for the point, line, or region between the reference image and the comparison image. In this way, it is possible to accurately extract correspondences in a widely separated range, and even when the user sets a search range, it is necessary to specify strict correspondence points only by specifying only approximate points. Does not improve operability
- FIG. 16 is a block diagram of a camera shake correction apparatus according to the present embodiment.
- the camera shake correction device 6 includes a motion vector calculation unit 13, a motion vector data storage unit 15, and an image composition that creates a composite image data by superimposing images using a plurality of images and their motion vectors.
- a means (function) 63, an image editing means (function) 64 for editing the composite image data, and an image data output means (function) 65 for outputting the created image data are provided.
- the image synthesis means 63 of the camera shake correction apparatus 6 inputs the input image data and the motion vector mesh data C stored in the motion vector data storage unit 15. Then, a conversion process is executed to match the comparison image with the reference image using the motion vector data.
- the motion vector in the case of camera shake has the same direction and size at each point, so the functions such as mesh designation means 33 and triangulation means 35 are deleted to simplify the function. You may make it let it.
- a reference image (standard image) is designated for two or more images, and an image is generated using a mesh-structured motion vector group from other images to the reference image. Since it is transformed and combined, the camera's CCD noise can be removed to create a clear image.
- FIG. 17 is a block diagram when the camera shake correction device 6 is incorporated in an imaging device such as a camera.
- an imaging device such as a camera.
- the motion vectorization calculation shown in Fig. 14 is highly practical as a simplified function.
- the image stabilization function can be realized by a microprocessor or an IC having an arithmetic function, and high-quality image data subjected to the image stabilization process can be obtained by being incorporated in the imaging apparatus. .
- FIG. 18 is a block diagram of the moving image generating apparatus according to the present embodiment.
- the moving image generating device 8 estimates a temporal change between the motion vector calculation unit 13, the motion vector data storage unit 15, and the motion vectors of the reference image (initial image) and the comparison image (final image).
- Intermediate data generation means (function) 83 for calculating intermediate motion vectors
- moving picture data generation means (function) 84 for creating video data from the intermediate motion vectors and input images.
- Movie data output means (function) 85 for output is provided.
- the intermediate data generation means 83 calculates the speed and acceleration of the change of the motion vector as a feature point or mesh data in the image, obtains a motion vector for each predetermined time segment, Save to intermediate data file 87.
- the moving image data generating means 84 arranges the input image data and the intermediate motion vector in a time series and stores them in the moving image data file 88.
- Fig. 19 shows an example of the data structure of a video data file. Each input image and the motion vector between them can be extracted in time series.
- the motion vector calculation method according to the present invention is applied to each of the above-described devices, and when the images are arranged in order and the objects in the images are moving according to a certain rule, the image- It can also be applied to matching processing for based VR (virtual reality).
- VR virtual reality
- the motion vector calculation method is an image stabilization apparatus for processing a camera shake image on a computer, an imaging apparatus such as a digital camera or a video camera equipped with a camera shake correction function, and accuracy with a small amount of data using motion vector data. It can be applied to a moving image generating apparatus that generates moving image data well.
- FIG. 1 is an explanatory diagram of motion vectors, reliability, and correlation according to the present invention.
- FIG. 2 is a block diagram of a motion vector computing device according to the first embodiment of the present invention.
- FIG. 3 Explanatory diagram of points, lines, and regions of motion vector mesh data
- Fig. 3 (a) is a sample image
- Fig. 3 (b) is an explanatory diagram of the correspondence between points, lines, and regions on the image It is.
- FIG. 4 Data structure diagram of motion vector mesh data.
- Fig. 4 (a) shows motion vector points.
- FIG. 4 (b) shows a motion vector line
- FIG. 4 (c) shows a data configuration example of a motion vector area.
- FIG. 5 is a flowchart showing a processing procedure of the multi-resolution generation means of FIG.
- FIG. 6 is an example of multi-resolution image data.
- FIG. 7 An illustration of how to specify motion vector mesh data A.
- Fig. 7 (a) shows the reference image and the comparison image
- Fig. 7 (b) shows the correspondence on the data between the two images.
- FIG. 8 is an explanatory diagram of image feature points.
- FIG. 9 is an explanatory diagram of motion vector mesh data B that is an output of the triangulation means of FIG. 2.
- FIG. 10 is a flowchart showing a processing procedure of the motion vector estimation means of FIG.
- FIG. 11 is a flowchart showing the matching processing procedure of FIG. (When there are two images)
- FIG. 12 is a flowchart showing the matching processing procedure of FIG. (If there are multiple images )
- FIG. 13 is a flowchart showing the smoothing processing procedure of FIG.
- FIG. 14 is a block diagram of another embodiment 1 according to the motion vector calculation apparatus of the present invention c
- FIG. 16 is a block diagram of a camera shake correction device according to a second embodiment of the present invention.
- FIG. 17 is a block diagram of an imaging apparatus according to a third embodiment of the present invention.
- FIG. 18 is a block diagram of a moving image generating apparatus according to a fourth embodiment of the present invention.
- FIG. 19 is a data configuration diagram of the moving image data file of FIG.
- Motion vector mesh data A file Image feature point data file Motion vector mesh data B file Motion vector mesh data C file Shake correction data calculation unit Shake correction data storage unit Image composition means
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
- Studio Devices (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Image Processing (AREA)
Abstract
Description
Claims
Priority Applications (7)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
AT05703663T ATE481696T1 (de) | 2005-01-14 | 2005-01-14 | Verfahren zur berechnung von bewegungsvektoren, vorrichtung zur korrektur von handbewegungen, die das verfahren verwendet, vorrichtung zur bildaufnahme und zur erzeugung von filmen |
DE200560023661 DE602005023661D1 (de) | 2005-01-14 | 2005-01-14 | Verfahren zur berechnung von bewegungsvektoren, vorrichtung zur korrektur von handbewegungen, die das verfahren verwendet, vorrichtung zur bildaufnahme und zur erzeugung von filmen |
EP20050703663 EP1843294B1 (en) | 2005-01-14 | 2005-01-14 | Motion vector calculation method, hand-movement correction device using the method, imaging device, and motion picture generation device |
KR20077015833A KR101036787B1 (ko) | 2005-01-14 | 2005-01-14 | 움직임 벡터 연산 방법과 이 방법을 이용한 손 떨림 보정장치, 촬상 장치, 및 동영상 생성 장치 |
PCT/JP2005/000424 WO2006075394A1 (ja) | 2005-01-14 | 2005-01-14 | 動きベクトル演算方法とこの方法を用いた手ぶれ補正装置、撮像装置、並びに動画生成装置 |
US11/795,218 US7847823B2 (en) | 2005-01-14 | 2005-01-14 | Motion vector calculation method and hand-movement correction device, imaging device and moving picture generation device |
JP2006552817A JP3935500B2 (ja) | 2005-01-14 | 2005-01-14 | 動きベクトル演算方法とこの方法を用いた手ぶれ補正装置、撮像装置、並びに動画生成装置 |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2005/000424 WO2006075394A1 (ja) | 2005-01-14 | 2005-01-14 | 動きベクトル演算方法とこの方法を用いた手ぶれ補正装置、撮像装置、並びに動画生成装置 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2006075394A1 true WO2006075394A1 (ja) | 2006-07-20 |
Family
ID=36677430
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2005/000424 WO2006075394A1 (ja) | 2005-01-14 | 2005-01-14 | 動きベクトル演算方法とこの方法を用いた手ぶれ補正装置、撮像装置、並びに動画生成装置 |
Country Status (7)
Country | Link |
---|---|
US (1) | US7847823B2 (ja) |
EP (1) | EP1843294B1 (ja) |
JP (1) | JP3935500B2 (ja) |
KR (1) | KR101036787B1 (ja) |
AT (1) | ATE481696T1 (ja) |
DE (1) | DE602005023661D1 (ja) |
WO (1) | WO2006075394A1 (ja) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2007226643A (ja) * | 2006-02-24 | 2007-09-06 | Morpho Inc | 画像処理装置 |
JP2008226187A (ja) * | 2007-03-15 | 2008-09-25 | Toshiba Corp | 動き推定装置及びその方法 |
JP2008282386A (ja) * | 2007-05-10 | 2008-11-20 | Honda Motor Co Ltd | 物体検出装置、物体検出方法及び物体検出プログラム |
EP2015560A2 (en) | 2007-07-13 | 2009-01-14 | Morpho Inc. | Image data processing method and imaging apparatus |
US20090028462A1 (en) * | 2007-07-26 | 2009-01-29 | Kensuke Habuka | Apparatus and program for producing a panoramic image |
JPWO2008032392A1 (ja) * | 2006-09-14 | 2010-01-21 | 富士通株式会社 | 画像処理方法および装置とそのプログラム |
JP2011504266A (ja) * | 2007-11-12 | 2011-02-03 | クゥアルコム・インコーポレイテッド | ブロックベースの画像安定化 |
JP2013190947A (ja) * | 2012-03-13 | 2013-09-26 | Morpho Inc | 画像処理装置、画像処理方法及び画像処理プログラム |
JP2017085598A (ja) * | 2016-12-08 | 2017-05-18 | 株式会社東芝 | 画像処理装置、画像処理方法、画像処理プログラム、および立体画像表示装置 |
US10003782B2 (en) | 2012-11-01 | 2018-06-19 | Kabushiki Kaisha Toshiba | Image processing device, method, computer-readable medium and 3D image display |
US10818018B2 (en) | 2016-11-24 | 2020-10-27 | Canon Kabushiki Kaisha | Image processing apparatus, image processing method, and non-transitory computer-readable storage medium |
CN113873099A (zh) * | 2021-08-27 | 2021-12-31 | 山东信通电子股份有限公司 | 一种输电通道视频稳像方法、设备及介质 |
CN114612773A (zh) * | 2022-02-25 | 2022-06-10 | 武汉大学 | 一种适用于sar和光学影像的高效海冰运动提取方法及系统 |
Families Citing this family (34)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7646916B2 (en) * | 2005-04-15 | 2010-01-12 | Mississippi State University | Linear analyst |
US8165205B2 (en) * | 2005-09-16 | 2012-04-24 | Sony Corporation | Natural shaped regions for motion compensation |
JP4959237B2 (ja) * | 2006-06-22 | 2012-06-20 | オリンパス株式会社 | 撮像システム及び撮像プログラム |
US8375302B2 (en) * | 2006-11-17 | 2013-02-12 | Microsoft Corporation | Example based video editing |
JP2008217526A (ja) * | 2007-03-06 | 2008-09-18 | Canon Inc | 画像処理装置、画像処理プログラム及び画像処理方法 |
US8773423B2 (en) * | 2007-05-07 | 2014-07-08 | Microsoft Corporation | Creating optimized gradient mesh of a vector-based image from a raster-based image |
JP4989308B2 (ja) * | 2007-05-16 | 2012-08-01 | キヤノン株式会社 | 画像処理装置及び画像検索方法 |
US7956899B2 (en) * | 2007-08-29 | 2011-06-07 | Sanyo Electric Co., Ltd. | Imaging device and image processing apparatus |
JP4940164B2 (ja) * | 2008-02-07 | 2012-05-30 | オリンパス株式会社 | 撮像装置及び撮像方法 |
JP5048542B2 (ja) * | 2008-02-07 | 2012-10-17 | オリンパス株式会社 | 画像処理装置及び画像処理プログラム |
US8379152B2 (en) | 2008-03-31 | 2013-02-19 | Sharp Laboratories Of America, Inc. | Systems and methods for increasing the temporal resolution of video data |
US8199243B2 (en) * | 2008-04-28 | 2012-06-12 | Panasonic Corporation | Imaging device and camera body |
JP4600530B2 (ja) * | 2008-06-17 | 2010-12-15 | ソニー株式会社 | 画像処理装置および画像処理方法、並びにプログラム |
JP4385077B1 (ja) | 2008-08-27 | 2009-12-16 | 三菱電機株式会社 | 動きベクトル検出装置および画像処理装置 |
CN102160381A (zh) * | 2008-09-24 | 2011-08-17 | 索尼公司 | 图像处理设备和方法 |
US8508659B2 (en) * | 2009-08-26 | 2013-08-13 | Nxp B.V. | System and method for frame rate conversion using multi-resolution temporal interpolation |
KR101451137B1 (ko) * | 2010-04-13 | 2014-10-15 | 삼성테크윈 주식회사 | 손떨림 검출 장치 및 방법 |
JP5445363B2 (ja) | 2010-07-08 | 2014-03-19 | 株式会社リコー | 画像処理装置、画像処理方法および画像処理プログラム |
JP5661359B2 (ja) * | 2010-07-16 | 2015-01-28 | キヤノン株式会社 | 画像処理装置、画像処理方法、およびプログラム |
JP5700968B2 (ja) | 2010-07-16 | 2015-04-15 | キヤノン株式会社 | 画像処理装置、画像処理方法、およびプログラム |
JP5791241B2 (ja) | 2010-07-16 | 2015-10-07 | キヤノン株式会社 | 画像処理方法、画像処理装置、およびプログラム |
JP5652649B2 (ja) * | 2010-10-07 | 2015-01-14 | 株式会社リコー | 画像処理装置、画像処理方法および画像処理プログラム |
EP2590140A1 (en) | 2011-09-05 | 2013-05-08 | Morpho, Inc. | Facial authentication system, facial authentication method, and facial authentication program |
US9202431B2 (en) * | 2012-10-17 | 2015-12-01 | Disney Enterprises, Inc. | Transfusive image manipulation |
KR101767927B1 (ko) | 2012-11-01 | 2017-08-17 | 한화테크윈 주식회사 | 실시간 움직임 검출 방법 및 시스템 |
US9374532B2 (en) | 2013-03-15 | 2016-06-21 | Google Inc. | Cascaded camera motion estimation, rolling shutter detection, and camera shake detection for video stabilization |
JP2014225108A (ja) * | 2013-05-16 | 2014-12-04 | ソニー株式会社 | 画像処理装置、画像処理方法およびプログラム |
JP6686890B2 (ja) * | 2014-10-03 | 2020-04-22 | 日本電気株式会社 | 情報処理装置、情報処理方法、及び、プログラム |
JP2016164709A (ja) | 2015-03-06 | 2016-09-08 | キヤノン株式会社 | 画像処理装置、撮像装置および画像処理プログラム |
US11272124B2 (en) | 2017-10-31 | 2022-03-08 | Morpho, Inc. | Image compositing device, image compositing method, and storage medium |
KR102042131B1 (ko) | 2018-01-30 | 2019-11-07 | 광운대학교 산학협력단 | 단말기에서 실시간 글자 인식시 영상을 안정화하는 방법 |
US10708597B2 (en) | 2018-02-01 | 2020-07-07 | Microsoft Technology Licensing, Llc | Techniques for extrapolating image frames |
KR102627646B1 (ko) * | 2018-10-23 | 2024-01-19 | 엘지전자 주식회사 | 신호 처리 장치 및 이를 구비하는 영상표시장치 |
WO2024049197A1 (ko) * | 2022-08-30 | 2024-03-07 | 엘지전자 주식회사 | 3d 데이터 송신 장치, 3d 데이터 송신 방법, 3d 데이터 수신 장치 및 3d 데이터 수신 방법 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH07154801A (ja) * | 1993-11-29 | 1995-06-16 | Ricoh Co Ltd | 階層型動きベクトル検出方法 |
JPH0973540A (ja) * | 1995-09-04 | 1997-03-18 | Sharp Corp | 動きベクトル算出装置 |
JPH11506576A (ja) * | 1995-03-18 | 1999-06-08 | テーウー エレクトロニクス カンパニー リミテッド | 特徴点ベース動き推定技法を用いる映像信号符号化方法及びその装置 |
JP2003078807A (ja) * | 2001-08-31 | 2003-03-14 | Sony Corp | 動きベクトル検出装置および方法、手振れ補正装置および方法、並びに撮像装置 |
Family Cites Families (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH06153146A (ja) | 1992-11-04 | 1994-05-31 | Matsushita Electric Ind Co Ltd | 動画像のシーンチェンジ検出装置および編集装置 |
JP3519441B2 (ja) | 1993-02-26 | 2004-04-12 | 株式会社東芝 | 動画像伝送装置 |
KR100188116B1 (ko) * | 1995-12-28 | 1999-06-01 | 김광호 | 손떨림 영상 안정화 회로 |
KR100265720B1 (ko) | 1997-03-31 | 2000-09-15 | 윤종용 | 2차원삼각형선격자모델을이용한동영상의움직임보상방법 |
EP1090502B1 (en) * | 1999-04-26 | 2005-11-30 | Koninklijke Philips Electronics N.V. | Sub-pixel accurate motion vector estimation and motion-compensated interpolation |
JP4296693B2 (ja) * | 2000-07-13 | 2009-07-15 | ソニー株式会社 | Av信号記録再生装置、cm検出方法、および記録媒体 |
KR20020076296A (ko) * | 2000-12-11 | 2002-10-09 | 코닌클리케 필립스 일렉트로닉스 엔.브이. | 비디오 신호 처리에서 모션 보상된 디-인터레이싱 |
KR100396558B1 (ko) * | 2001-10-25 | 2003-09-02 | 삼성전자주식회사 | 적응 움직임 보상형 프레임 및/또는 레이트 변환 장치 및그 방법 |
JP3804617B2 (ja) * | 2003-02-14 | 2006-08-02 | コニカミノルタフォトイメージング株式会社 | 画像処理装置及び方法 |
US7499494B2 (en) * | 2003-12-23 | 2009-03-03 | Genesis Microchip Inc. | Vector selection decision for pixel interpolation |
US7433497B2 (en) * | 2004-01-23 | 2008-10-07 | Hewlett-Packard Development Company, L.P. | Stabilizing a sequence of image frames |
US7616782B2 (en) * | 2004-05-07 | 2009-11-10 | Intelliview Technologies Inc. | Mesh based frame processing and applications |
-
2005
- 2005-01-14 JP JP2006552817A patent/JP3935500B2/ja active Active
- 2005-01-14 EP EP20050703663 patent/EP1843294B1/en not_active Not-in-force
- 2005-01-14 AT AT05703663T patent/ATE481696T1/de not_active IP Right Cessation
- 2005-01-14 KR KR20077015833A patent/KR101036787B1/ko active IP Right Grant
- 2005-01-14 US US11/795,218 patent/US7847823B2/en active Active
- 2005-01-14 WO PCT/JP2005/000424 patent/WO2006075394A1/ja active Application Filing
- 2005-01-14 DE DE200560023661 patent/DE602005023661D1/de active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH07154801A (ja) * | 1993-11-29 | 1995-06-16 | Ricoh Co Ltd | 階層型動きベクトル検出方法 |
JPH11506576A (ja) * | 1995-03-18 | 1999-06-08 | テーウー エレクトロニクス カンパニー リミテッド | 特徴点ベース動き推定技法を用いる映像信号符号化方法及びその装置 |
JPH0973540A (ja) * | 1995-09-04 | 1997-03-18 | Sharp Corp | 動きベクトル算出装置 |
JP2003078807A (ja) * | 2001-08-31 | 2003-03-14 | Sony Corp | 動きベクトル検出装置および方法、手振れ補正装置および方法、並びに撮像装置 |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2007226643A (ja) * | 2006-02-24 | 2007-09-06 | Morpho Inc | 画像処理装置 |
JP4750854B2 (ja) * | 2006-09-14 | 2011-08-17 | 富士通株式会社 | 画像処理方法および装置とそのプログラム |
US8311367B2 (en) | 2006-09-14 | 2012-11-13 | Fujitsu Limited | Image processing device |
JPWO2008032392A1 (ja) * | 2006-09-14 | 2010-01-21 | 富士通株式会社 | 画像処理方法および装置とそのプログラム |
JP2008226187A (ja) * | 2007-03-15 | 2008-09-25 | Toshiba Corp | 動き推定装置及びその方法 |
JP2008282386A (ja) * | 2007-05-10 | 2008-11-20 | Honda Motor Co Ltd | 物体検出装置、物体検出方法及び物体検出プログラム |
EP2015560A2 (en) | 2007-07-13 | 2009-01-14 | Morpho Inc. | Image data processing method and imaging apparatus |
US20090028462A1 (en) * | 2007-07-26 | 2009-01-29 | Kensuke Habuka | Apparatus and program for producing a panoramic image |
EP2023596A2 (en) | 2007-07-26 | 2009-02-11 | Morpho Inc. | Apparatus and program for producing as panoramic image |
US8588546B2 (en) * | 2007-07-26 | 2013-11-19 | Morpho, Inc. | Apparatus and program for producing a panoramic image |
JP2011504266A (ja) * | 2007-11-12 | 2011-02-03 | クゥアルコム・インコーポレイテッド | ブロックベースの画像安定化 |
US8600189B2 (en) | 2007-11-12 | 2013-12-03 | Qualcomm Incorporated | Block-based image stabilization |
JP2013190947A (ja) * | 2012-03-13 | 2013-09-26 | Morpho Inc | 画像処理装置、画像処理方法及び画像処理プログラム |
US10003782B2 (en) | 2012-11-01 | 2018-06-19 | Kabushiki Kaisha Toshiba | Image processing device, method, computer-readable medium and 3D image display |
US10818018B2 (en) | 2016-11-24 | 2020-10-27 | Canon Kabushiki Kaisha | Image processing apparatus, image processing method, and non-transitory computer-readable storage medium |
JP2017085598A (ja) * | 2016-12-08 | 2017-05-18 | 株式会社東芝 | 画像処理装置、画像処理方法、画像処理プログラム、および立体画像表示装置 |
CN113873099A (zh) * | 2021-08-27 | 2021-12-31 | 山东信通电子股份有限公司 | 一种输电通道视频稳像方法、设备及介质 |
CN113873099B (zh) * | 2021-08-27 | 2024-04-12 | 山东信通电子股份有限公司 | 一种输电通道视频稳像方法、设备及介质 |
CN114612773A (zh) * | 2022-02-25 | 2022-06-10 | 武汉大学 | 一种适用于sar和光学影像的高效海冰运动提取方法及系统 |
Also Published As
Publication number | Publication date |
---|---|
EP1843294B1 (en) | 2010-09-15 |
EP1843294A1 (en) | 2007-10-10 |
EP1843294A4 (en) | 2009-06-03 |
DE602005023661D1 (de) | 2010-10-28 |
US7847823B2 (en) | 2010-12-07 |
JPWO2006075394A1 (ja) | 2008-06-12 |
JP3935500B2 (ja) | 2007-06-20 |
KR20070093995A (ko) | 2007-09-19 |
US20080180535A1 (en) | 2008-07-31 |
KR101036787B1 (ko) | 2011-05-25 |
ATE481696T1 (de) | 2010-10-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP3935500B2 (ja) | 動きベクトル演算方法とこの方法を用いた手ぶれ補正装置、撮像装置、並びに動画生成装置 | |
JP4620607B2 (ja) | 画像処理装置 | |
WO2021088473A1 (en) | Image super-resolution reconstruction method, image super-resolution reconstruction apparatus, and computer-readable storage medium | |
US9609181B2 (en) | Image signal processor and method for synthesizing super-resolution images from non-linear distorted images | |
US20200258196A1 (en) | Image processing apparatus, image processing method, and storage medium | |
CN108463994B (zh) | 图像处理装置、图像处理方法和存储介质 | |
US20140072232A1 (en) | Super-resolution method and apparatus for video image | |
CN107610153B (zh) | 电子设备以及相机 | |
WO2013131929A1 (en) | Method and apparatus for performing super-resolution | |
JP2009033392A (ja) | パノラマ画像生成装置およびプログラム | |
US9070223B2 (en) | Image processing device, image processing method, and image processing program | |
WO2012169174A1 (ja) | 画像処理装置および画像処理方法 | |
JP5820716B2 (ja) | 画像処理装置、画像処理方法、コンピュータプログラム、記録媒体、立体画像表示装置 | |
US7522189B2 (en) | Automatic stabilization control apparatus, automatic stabilization control method, and computer readable recording medium having automatic stabilization control program recorded thereon | |
CN112929562B (zh) | 视频抖动的处理方法、装置、设备以及存储介质 | |
JP4212430B2 (ja) | 多重画像作成装置、多重画像作成方法、多重画像作成プログラム及びプログラム記録媒体 | |
JP6600113B1 (ja) | Ar用の画像処理装置、画像処理システム及びプログラム | |
JP5713256B2 (ja) | 画像処理装置、撮像装置、および画像処理プログラム | |
JP4858908B2 (ja) | 画像処理方法および撮像装置 | |
RU2576490C1 (ru) | Способ гибридного ретуширования фона для преобразования 2d в 3d | |
JP2024007899A (ja) | 画像処理装置、画像処理方法、およびプログラム | |
JP2014060667A (ja) | 画像処理装置、画像処理システム、画像処理方法及びプログラム |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 2006552817 Country of ref document: JP |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2005703663 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1020077015833 Country of ref document: KR |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWP | Wipo information: published in national office |
Ref document number: 2005703663 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 11795218 Country of ref document: US |