WO2011030399A1 - 画像処理方法及び装置 - Google Patents
画像処理方法及び装置 Download PDFInfo
- Publication number
- WO2011030399A1 WO2011030399A1 PCT/JP2009/065683 JP2009065683W WO2011030399A1 WO 2011030399 A1 WO2011030399 A1 WO 2011030399A1 JP 2009065683 W JP2009065683 W JP 2009065683W WO 2011030399 A1 WO2011030399 A1 WO 2011030399A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- vector
- depth
- innermost
- parallax
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/97—Determining parameters from multiple pictures
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
- H04N13/20—Image signal generators
- H04N13/204—Image signal generators using stereoscopic image cameras
- H04N13/207—Image signal generators using stereoscopic image cameras using a single 2D image sensor
- H04N13/221—Image signal generators using stereoscopic image cameras using a single 2D image sensor using the relative movement between cameras and objects
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
- H04N13/20—Image signal generators
- H04N13/261—Image signal generators with monoscopic-to-stereoscopic image conversion
- H04N13/264—Image signal generators with monoscopic-to-stereoscopic image conversion using the relative movement of objects in two video frames or fields
Definitions
- the present invention relates to an image processing method and apparatus for generating a parallax image for stereoscopic display from a two-dimensional image.
- a stereoscopic display technique for recognizing a stereoscopic image by displaying each parallax image so as to be visually recognized by the left eye and the right eye is known. Due to the triangulation procedure, the image when viewed with the left eye is slightly displaced when viewed with the right eye. By grasping the deviation in the manner of triangulation, it is determined whether this object is in front or behind. That is, to realize stereoscopic display, a three-dimensional image having binocular parallax between images is required.
- Patent Document 1 detects a motion vector from a reference frame of another two-dimensional moving image to another frame, determines a frame delay amount based on the horizontal component of the motion vector, and sets a line based on the vertical component of the motion vector.
- a method of determining a delay amount, generating a parallax image corresponding to a reference frame from the frame / line delay amount, and stereoscopically viewing the parallax image based on the principle of motion stereo it is possible to correct the vertical movement of the entire screen and generate a correct parallax image.
- the object is blurred. Therefore, a correct parallax image cannot be generated.
- Patent Document 2 a motion vector and a background region are detected, a background vector is calculated from the motion vector of the background region, a relative vector is calculated by subtracting the background vector from all motion vectors, and the magnitude of the relative vector is There is provided a method for generating a parallax image so that a larger one is arranged in front.
- this method depending on the detection result of the background region, there may be a parallax image with an uncomfortable feeling that an object existing in the back is arranged in front.
- One aspect of the present invention is an image processing method for generating a parallax image for three-dimensional display from a two-dimensional moving image, a first image at an arbitrary time of the two-dimensional moving image, and a time different from the first image And detecting a motion vector between the second image and the second image for each block of the first image based on the second image; Detecting a difference vector between each motion vector and the innermost vector, giving a nearer depth to the block of the first image corresponding to the motion vector having a larger difference vector, And a step of generating one or more parallax images from the depth and the depth.
- FIG. 1 is a block diagram of an image processing apparatus that generates a parallax image for stereoscopic display from a two-dimensional image according to a first embodiment of the present invention. It is a flowchart figure which shows the image processing method which concerns on the 1st Embodiment of this invention. It is a figure for demonstrating the principle of motion stereo. It is a figure for demonstrating that a motion blurs. It is a figure for demonstrating the conversion model from a parallax vector to depth. It is a figure for demonstrating the conflict of the object of an image. It is a flowchart figure which shows the innermost vector detection method. It is a figure for demonstrating a depth conversion model. It is a figure for demonstrating parallax image generation.
- the frame memory 11 stores an input moving image from a video apparatus such as a video camera or a video player in units of frames.
- This frame memory 11 is connected to a motion vector detector 21.
- the motion vector detector 21 detects a motion vector from the input moving image and the image stored in the frame memory 11 by a known method such as block matching.
- the output of the motion vector detector 21 is connected to the clustering device 23.
- the clustering device 23 divides the image (frame) into a plurality of regions (clusters).
- the output of the clustering device 23 is connected to the innermost vector detector 24.
- the innermost vector detector 24 detects the vector of the innermost object, that is, the innermost vector based on the cluster.
- the overlap between the object in the back and the object in the foreground is detected based on the cluster, the most overlapping cluster is determined as the innermost, and the vector of this cluster is detected as the innermost vector.
- the output of the innermost vector detector 24 is connected to a depth calculator 25, which calculates the depth from the innermost vector. Specifically, the depth from the backmost vector and the vector of the foremost object, that is, the depth from the frontmost vector to the backmost object is calculated.
- the output of the depth calculator 25 is connected to a disparity vector calculator 26, and the disparity vector calculator 26 calculates a disparity vector based on the depth information.
- the output of the parallax vector calculator 26 is connected to a parallax image generator 27.
- the parallax image generator 27 creates a video viewed with the right eye and a video viewed with the left eye from the parallax vector by a method described later.
- the basis for generating a three-dimensional image from a two-dimensional image uses the principle of motion stereo as shown in FIG.
- the video at time t1 is very similar to the image seen with the left eye
- the video at time t2 is very similar to the image seen with the right eye.
- Stereoscopic viewing is possible by viewing the video of such a moving camera with the right and left eyes.
- motion stereo is a technique that uses moving images from a moving camera as a left parallax image and a right parallax image, respectively.
- a moving camera as a left parallax image and a right parallax image, respectively.
- the images at different times will enter the right and left eyes, and the hands and feet will be shaken and the object will be correctly It cannot be stereoscopically viewed.
- a motion vector from the reference frame of the input moving image to a frame at another time is obtained, the motion vector is regarded as a disparity vector, a depth map is calculated from the disparity vector, and the calculated depth map is used again.
- a disparity vector is calculated, and two disparity images are generated according to the disparity vector from the reference frame (however, the number is not limited to two). In this way, a parallax image at the same time can be generated from one (one time) reference frame, and thus the above-described problem of hand and foot shake is solved.
- the horizontal component of a motion vector is a disparity vector
- the conversion from the disparity vector to the depth follows the model shown in FIG. 5 from the geometric relationship between the human right and left eyes and the target object. Since the left and right eyes are arranged in the horizontal direction, the parallax basically occurs in the horizontal direction. According to this model, it is possible to reproduce the natural depth from the movement.
- the video is limited to a video having a horizontal motion.
- a method of tracking the depth from the motion of the moving image is used instead of detecting the background vector. That is, the motion vector (the deepest vector) of the area to be arranged on the farthest side is detected, and an object having a larger difference vector from the deepest vector is arranged in front. The innermost vector is detected from the overlapping relationship of motion vectors. In this way, it is possible to sequentially arrange the objects in front of the innermost area from the innermost area according to the difference of the motion vectors.
- the motion vector detector 22 detects a motion vector from the t frame to the t-1 frame.
- Various methods can be used to detect a motion vector.
- Block matching is a method in which a t frame is divided into rectangular block parts and a block corresponding to each block is searched from the t ⁇ 1 frame.
- the block size be M 1 and M 2 and the block position be i, j.
- Mean absolute difference (MAD) or the like can be used as an error function for obtaining motion.
- mean squared error may be used. If the search range is a rectangular area from ⁇ W to W, the block matching algorithm for obtaining the motion vector u (i, j) at the i, j position is as follows.
- the motion vector in the block is the same as the motion vector of the block.
- a motion vector used for compression in moving picture coding such as MPEG2
- a motion vector decoded by a decoder may be used. it can.
- ⁇ Clustering step S12> the clustering device 23 classifies objects having similar motion vector directions and magnitudes.
- a K-means method or the like can be used as a clustering method, but is not limited to this. The outline of processing by the K-means method will be described below.
- Step 3 For each motion vector, select a cluster having the smallest difference between the intra-cluster average vector and the motion vector (difference vector) in all clusters, and update the label with the cluster number.
- Step 4 Repeat steps 2 and 3 until the specified iteration.
- the motion vectors can be classified into K clusters, and K average vectors within the clusters are calculated.
- the innermost vector detector 24 selects a motion vector to be arranged on the innermost side from the detected average vector in the cluster.
- the innermost region can be said to be the region having the highest probability of being covered with other objects in the overlap between the objects. Therefore, such a region is determined from the overlap of motion vectors.
- FIG. 1 A cross-sectional view of the object moving on the background is shown in FIG.
- an area surrounded by a broken line is defined as an area “conflict” in which motion vectors collide with each other. This conflict indicates overlapping motion vectors. Attention is paid to the two vectors in the following equation (5) which are in conflict.
- the pixel difference value is smaller than the motion vector in the back. Therefore, the anteroposterior relationship of the two vectors conflicted by the absolute value pixel difference of the following equation can be determined.
- the cluster to which u (x 1 , y 1 ) and u (x 2 , y 2 ) belong can be identified by the labels l (x 1 , y 1 ) and l (x 2 , y 2 ). From the conflict of motion vectors, the context between clusters can be understood. This determination is performed on the entire screen to determine the innermost cluster.
- step S21 a motion vector conflict is detected (step S21).
- the context of the detected conflicting vector is determined (S22).
- the context of the corresponding cluster is updated (S23). It is determined whether the entire screen has been processed (S24). If this determination is Yes, the cluster that is on the farthest side is selected (S25), and the process is terminated. If the determination in step S24 is No, the process returns to step S21.
- ⁇ Depth calculation step S14> Let the average motion vector of the cluster be the innermost vector u_deep.
- the depth calculator 25 calculates a depth value from the motion vector and the innermost vector. As shown in FIG. 5, the depth value is obtained by using the similarity of two triangles, a triangle connecting the right eye, the left eye and the object, and a right parallax on the screen, a left parallax, and a triangle defined by the object. Calculated from a vector.
- each parameter shown in FIG. 5 is defined as follows.
- ⁇ ⁇ ⁇ is an operator for calculating the l2 norm (length) of the vector.
- the unit is a pixel, and the conversion from the pixel size to [cm] is performed according to the following equation (9).
- the depth value z is in the range of 0-255 (0-1 may be used), with 0 representing the front and 255 representing the back. However, this value is only virtual and needs to be converted into an actual distance. Conversion is performed according to the following equation (10) using the depth size L z in the real space.
- FIG. 8 shows a curve created by plotting the result calculated by Equation 12.
- the vertical axis of the graph in FIG. 8 indicates parallax, and the horizontal axis indicates depth.
- the above-mentioned depth conversion model makes the motion vector larger in the right direction deeper and the motion vector larger in the left direction closer.
- the axis is reversed, the opposite is true.
- the above-described depth conversion model is an inversely proportional model to the disparity vector, but the depth conversion model may be, for example, a function model that approximates an inversely proportional relationship by a partial proportional relationship.
- the distance z s to the screen is a value directly related to the calculated depth value, and the distance z s to the screen may be determined so that the histogram of the calculated depth value is as wide as possible.
- ⁇ Parallax vector calculation step S15> Once the depth value is calculated, a parallax image can be generated according to an arbitrary parameter.
- the disparity vector calculator 26 calculates a disparity vector from which a disparity image is generated from the depth value.
- Stereoscopic parameters b, z s , z 0 and L z can be arbitrarily determined based on the stereoscopic vision desired to be provided. For example, z s is determined according to the actual screen position, and z 0 is increased when it is desired to increase the pop-out. The depth of the depth can be determined by L z.
- the disparity vector can be calculated from the depth value according to the depth disparity vector conversion model of the following equation (13) obtained by modifying the conversion model of equation (2).
- parallax image generation step S16 parallax images are generated from t frames and parallax vectors for the number of parallax images that the parallax image generator 27 wants to generate.
- the left parallax image can be generated by moving the pixel value I t (x, y) of t frame according to d L.
- the case of two parallaxes has been described as an example, but the same processing may be performed for multi-parallax cases.
- the left parallax image and the right parallax image generated as described above are alternately switched at high speed or temporally or spatially in such a way that the first line on the screen is viewed with the left eye and the second line is viewed with the right eye.
- the display is displayed on the display as a single screen so that the left and right eyes can see different images.
- images at different times are stored in the frame memory, and motion vectors are detected for the entire screen from the images at different times.
- This motion vector is divided into vectors having the same direction and collected. That is, clustering is performed.
- clustering a plurality of clusters are extracted from the screen. By looking at the overlap of these clusters, the most covered cluster is detected. Let the average motion vector of the most covered cluster be the innermost vector.
- the motion vector of each cluster is subtracted from the innermost vector, and the larger the difference vector is, the more the difference vector is smaller in the foreground, and it is determined that the depth vector is in the back.
- left and right parallax images are generated for each object.
- the method of the present invention described in the embodiment of the present invention can be executed by a computer, and as a program that can be executed by the computer, a magnetic disk (flexible disk, hard disk, etc.), an optical disk (CD-ROM) , DVD, etc.) and storage media such as semiconductor memory can also be distributed.
- a magnetic disk flexible disk, hard disk, etc.
- an optical disk CD-ROM
- DVD digital versatile disk
- storage media such as semiconductor memory
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
- Image Analysis (AREA)
Abstract
Description
第1実施形態に従った2次元画像から立体表示のための視差画像を生成する画像処理装置を図1のブロック図を参照して説明する。
2次元の画像から3次元の画像を生成する根拠は図3に示すようなモーションステレオの原理を利用する。すなわち1台のカメラが左から右に移動しながら撮影している状況を考えると、t1の時刻の映像は左目で見える画像に、t2の時刻の映像は右目で見える画像に酷似していると考えることができる。このような移動カメラの映像を右目及び左目で見ることにより立体視が可能となる。このように移動カメラからの動画像をそれぞれ左視差画像及び右視差画像として利用する技術がモーションステレオである。しかし図4に示すように、画面内にカメラの動きとは独立して動いている物体があると異なる時刻の映像が右目と左目に入ることになり、手や足がぶれてしまい物体を正しく立体視することができない。
動きベクトル検出ステップS11では、動きベクトル検出器22がtフレームからt-1フレームへの動きベクトルを検出する。動きベクトルの検出には様々な方法が使えるが、ここではブロックマッチングを使ったものを述べる。ただしそれに限ったものではない。ブロックマッチングはtフレームを矩形ブロックの部分に分割し、ブロック毎に対応するブロックをt-1フレーム上から探索する方法である。ブロックの大きさをM1,M2とし、ブロックの位置をi,jとする。動きを求めるための誤差関数として平均絶対値差分(Mean Absolute Difference:MAD)などを用いることができる。
クラスタリングステップS12では、クラスタリング装置23が、動きベクトルの方向及び大きさが近い対象同士を分類する。クラスタリング手法としてはK-means法などを用いることができるが、これに限ったものではない。以下K-means法による処理の概要を述べる。
最奥ベクトル検出ステップS13では、最奥ベクトル検出器24が、最も奥側に配置すべき動きベクトルを、検出されたクラスタ内平均ベクトルの中から選択する。最も奥側の領域は物体間の重なりにおいて、他の物体から覆われている確率が最も高い領域ということができる。そこでそのような領域を動きベクトルの重なり合いから判断する。
クラスタの平均動きベクトルを最奥ベクトルu_deepとする。奥行き算出ステップS14では、奥行き算出器25が動きベクトルと最奥ベクトルから奥行き値を算出する。奥行き値は、図5に示すように、右目、左目及び物体を結ぶ三角形と、画面上での右視差、左視差及び物体により規定される三角形との二つの三角形の相似性を利用して視差ベクトルから算出される。ここで図5に示される各パラメータは次のように定義するものとする。
視差ベクトル:d[cm]
眼間距離:b[cm]
画面までの距離:zs[cm]
飛び出し距離:z0[cm]
奥行きの最大値:zmax
実空間での奥行きサイズ:Lz[cm]
ここでb、zs、z0、Lzは任意の値を設定しておく。視差は動きベクトルと最奥ベクトルから次式(8)によって計算できる。
一度奥行き値を算出しておけば、任意のパラメータに従って視差画像を生成できる。視差ベクトル算出器26は視差画像を生成する元となる視差ベクトルを奥行き値から算出する。
Claims (6)
- 2次元動画像から3次元表示のための視差画像を生成する画像処理方法において、
2次元動画像の任意時刻の第1画像と、前記第1画像とは別の時刻の第2画像に基づいて、前記第1画像のブロック毎に、第2画像との間の動きベクトルを検出するステップと、
奥行きを最も奥側とすべきブロックに対して求めた最奥ベクトルを、前記動きベクトルの中から検出するステップと、
各動きベクトルと前記最奥ベクトルとの差分ベクトルを求め、前記差分ベクトルが大きい前記動きベクトルに対応する前記第1画像のブロックほど手前側の奥行きを与えるステップと、
前記第1画像と前記奥行きから1枚以上の視差画像を生成するステップと、
を有する画像処理方法。 - 前記最奥ベクトルを検出するステップは、
前記動きベクトルを1つ以上のクラスタに分類し、前記クラスタ内の平均動きベクトルを求めるステップと、
前記平均動きベクトルのどれか1つを前記最奥ベクトルに設定するステップと、を含むことを特徴とする請求項1に記載の画像処理方法。 - 前記最奥ベクトル検出ステップは、
前記クラスタ間の動きベクトルの重なりを検出し、もっとも覆い被されているクラスタ内の平均動きベクトルを前記最奥ベクトルに設定するステップを含むことを特徴とする請求項2に記載の画像処理方法。 - 前記差分ベクトルの大きさと前記奥行きの関係が反比例していることを特徴とする請求項3に記載の画像処理方法。
- 前記差分ベクトルの大きさと前記奥行きの関係が比例していることを特徴とする請求項3に記載の画像処理方法。
- 2次元動画像から3次元表示のための視差画像を生成する画像処理装置において、
2次元動画像の任意時刻の第1画像と、前記第1画像とは別の時刻の第2画像に基づいて、前記第1画像のブロック毎に、第2画像との間の動きベクトルを検出する動きベクトル検出器と、
奥行きが奥側となる部分の最奥ベクトルを、前記動きベクトルの中から検出する奥行きベクトル検出器と、
各動きベクトルと前記最奥ベクトルとの差分ベクトルを求め、前記差分ベクトルが大きい前記動きベクトルに対応する前記第1画像のブロックほど手前側の奥行きを与える視差ベクトル算出器と、
前記第1画像と前記奥行きから1枚以上の視差画像を生成する視差画像生成器と、
を含む画像処理装置。
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2011501034A JP4892113B2 (ja) | 2009-09-08 | 2009-09-08 | 画像処理方法及び装置 |
PCT/JP2009/065683 WO2011030399A1 (ja) | 2009-09-08 | 2009-09-08 | 画像処理方法及び装置 |
CN200980157191.XA CN102326394B (zh) | 2009-09-08 | 2009-09-08 | 图像处理方法以及装置 |
US13/415,381 US8289376B2 (en) | 2009-09-08 | 2012-03-08 | Image processing method and apparatus |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2009/065683 WO2011030399A1 (ja) | 2009-09-08 | 2009-09-08 | 画像処理方法及び装置 |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US13/415,381 Continuation US8289376B2 (en) | 2009-09-08 | 2012-03-08 | Image processing method and apparatus |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2011030399A1 true WO2011030399A1 (ja) | 2011-03-17 |
Family
ID=43732087
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2009/065683 WO2011030399A1 (ja) | 2009-09-08 | 2009-09-08 | 画像処理方法及び装置 |
Country Status (4)
Country | Link |
---|---|
US (1) | US8289376B2 (ja) |
JP (1) | JP4892113B2 (ja) |
CN (1) | CN102326394B (ja) |
WO (1) | WO2011030399A1 (ja) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2014239436A (ja) * | 2013-06-06 | 2014-12-18 | ソニー株式会社 | 2次元コンテンツの3次元コンテンツへのリアルタイム変換の方法及び装置 |
JP2015512196A (ja) * | 2012-02-13 | 2015-04-23 | トムソン ライセンシングThomson Licensing | 3dグラフィックス・アニメーションを3dステレオ・コンテンツに挿入する方法および装置 |
US9483836B2 (en) | 2011-02-28 | 2016-11-01 | Sony Corporation | Method and apparatus for real-time conversion of 2-dimensional content to 3-dimensional content |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP5627498B2 (ja) * | 2010-07-08 | 2014-11-19 | 株式会社東芝 | 立体画像生成装置及び方法 |
JP2012151663A (ja) * | 2011-01-19 | 2012-08-09 | Toshiba Corp | 立体音響生成装置及び立体音響生成方法 |
JP6412819B2 (ja) * | 2015-03-30 | 2018-10-24 | 株式会社東芝 | 画像処理装置、画像処理方法、および画像処理プログラム |
JP6207640B2 (ja) * | 2016-01-27 | 2017-10-04 | エフ・エーシステムエンジニアリング株式会社 | 2次元映像の立体映像化表示装置 |
CN109829393B (zh) * | 2019-01-14 | 2022-09-13 | 北京鑫洋泉电子科技有限公司 | 一种移动物体探测方法、装置及存储介质 |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH07264631A (ja) * | 1994-03-24 | 1995-10-13 | Sanyo Electric Co Ltd | 2次元画像を3次元画像に変換する方法及び装置 |
JP2000261828A (ja) * | 1999-03-04 | 2000-09-22 | Toshiba Corp | 立体映像生成方法 |
JP2007502454A (ja) * | 2003-08-05 | 2007-02-08 | コーニンクレッカ フィリップス エレクトロニクス エヌ ヴィ | マルチビュー画像の生成 |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5259040A (en) * | 1991-10-04 | 1993-11-02 | David Sarnoff Research Center, Inc. | Method for determining sensor motion and scene structure and image processing system therefor |
-
2009
- 2009-09-08 WO PCT/JP2009/065683 patent/WO2011030399A1/ja active Application Filing
- 2009-09-08 CN CN200980157191.XA patent/CN102326394B/zh not_active Expired - Fee Related
- 2009-09-08 JP JP2011501034A patent/JP4892113B2/ja not_active Expired - Fee Related
-
2012
- 2012-03-08 US US13/415,381 patent/US8289376B2/en not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH07264631A (ja) * | 1994-03-24 | 1995-10-13 | Sanyo Electric Co Ltd | 2次元画像を3次元画像に変換する方法及び装置 |
JP2000261828A (ja) * | 1999-03-04 | 2000-09-22 | Toshiba Corp | 立体映像生成方法 |
JP2007502454A (ja) * | 2003-08-05 | 2007-02-08 | コーニンクレッカ フィリップス エレクトロニクス エヌ ヴィ | マルチビュー画像の生成 |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9483836B2 (en) | 2011-02-28 | 2016-11-01 | Sony Corporation | Method and apparatus for real-time conversion of 2-dimensional content to 3-dimensional content |
JP2015512196A (ja) * | 2012-02-13 | 2015-04-23 | トムソン ライセンシングThomson Licensing | 3dグラフィックス・アニメーションを3dステレオ・コンテンツに挿入する方法および装置 |
JP2014239436A (ja) * | 2013-06-06 | 2014-12-18 | ソニー株式会社 | 2次元コンテンツの3次元コンテンツへのリアルタイム変換の方法及び装置 |
Also Published As
Publication number | Publication date |
---|---|
JP4892113B2 (ja) | 2012-03-07 |
US8289376B2 (en) | 2012-10-16 |
CN102326394A (zh) | 2012-01-18 |
CN102326394B (zh) | 2014-06-11 |
JPWO2011030399A1 (ja) | 2013-02-04 |
US20120169844A1 (en) | 2012-07-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP4892113B2 (ja) | 画像処理方法及び装置 | |
JP5153940B2 (ja) | 動き補償を用いた画像の奥行き抽出のためのシステムおよび方法 | |
US9041819B2 (en) | Method for stabilizing a digital video | |
US9237330B2 (en) | Forming a stereoscopic video | |
EP2618584B1 (en) | Stereoscopic video creation device and stereoscopic video creation method | |
JP5156837B2 (ja) | 領域ベースのフィルタリングを使用する奥行マップ抽出のためのシステムおよび方法 | |
US8488869B2 (en) | Image processing method and apparatus | |
JP5561781B2 (ja) | 2d画像データを立体画像データに変換する方法およびシステム | |
US8611642B2 (en) | Forming a steroscopic image using range map | |
JP4762994B2 (ja) | 視差マップ | |
KR101502362B1 (ko) | 영상처리 장치 및 방법 | |
US20130127988A1 (en) | Modifying the viewpoint of a digital image | |
JP5425897B2 (ja) | 前方及び後方の深度予測を伴う、画像の深度抽出のシステム及び方法 | |
US20130129192A1 (en) | Range map determination for a video frame | |
JP2013527646A5 (ja) | ||
JP6173218B2 (ja) | 背景ピクセル拡張及び背景優先パッチマッチングを用いるマルチビューレンダリング装置及び方法 | |
KR20210038659A (ko) | 2차원 이미지를 외삽하기 위해 깊이 정보를 사용하는 시스템 및 방법 | |
US8976171B2 (en) | Depth estimation data generating apparatus, depth estimation data generating method, and depth estimation data generating program, and pseudo three-dimensional image generating apparatus, pseudo three-dimensional image generating method, and pseudo three-dimensional image generating program | |
WO2008152607A1 (en) | Method, apparatus, system and computer program product for depth-related information propagation | |
JP5127973B1 (ja) | 映像処理装置、映像処理方法および映像表示装置 | |
JP2014072809A (ja) | 画像生成装置、画像生成方法、画像生成装置用プログラム | |
JP6685241B2 (ja) | 立体視ビデオ生成 | |
US20130229408A1 (en) | Apparatus and method for efficient viewer-centric depth adjustment based on virtual fronto-parallel planar projection in stereoscopic images | |
KR101939243B1 (ko) | 입체 깊이 조정 및 초점 조정 | |
KR20170035885A (ko) | 입체 비디오 주밍 및 비디오 내의 전경 및 배경 검출 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 200980157191.X Country of ref document: CN |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2011501034 Country of ref document: JP |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 09849179 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 09849179 Country of ref document: EP Kind code of ref document: A1 |