Abstract:
3차원 메쉬 부호화 방법 및 장치가 개시된다. 3차원 메쉬 부호화 방법은, 3차원 객체에 대응하는 3차원 메쉬를 구성하는 게이트의 우선 순위를 결정하고, 결정된 게이트의 우선 순위를 이용하여 3차원 메쉬를 구성하는 정점들을 제거함으로써 3차원 메쉬를 단순화할 수 있다.
Abstract:
본 발명은 동영상에서 배경과 사용자 객체의 경계 추정 방법에 있어서, 객체 분할을 위한 동영상의 첫 프레임에서 초기 객체 경계를 추정하여 객체와 배경을 분할하는 과정과, 첫 프레임에서 분할된 객체와 배경의 컬러(color) 확률 밀도 함수를 각각 추정하는 과정과, 현재 프레임에서, 이전 프레임의 객체 경계를 따라 상기 이전 프레임의 객체 경계를 중심으로 미리 설정된 일정한 너비를 가지는 경계 스트립(strip)을 생성하여 상기 경계 스트립의 일관성(coherence) 확률 밀도 함수를 추정하는 과정과, 에너지 최소화(Energy Minimization)를 수행하여 현재 프레임의 객체를 분할하는 과정을 포함한다. 동영상, 화상 통신, 객체 경계 추정, 확률 밀도 함수, 경계 스트립
Abstract:
PURPOSE: A 3D mesh compressing device and method thereof are provided to improve a compression efficiency of whole 3D graphic data by reducing probability distribution area of difference vector when estimating geometry information of 3D graphic data. CONSTITUTION: A mesh simplification unit(110) simplifies an inputted 3D mesh to generates a base mesh. A base mesh encoding unit(120) encodes the base mesh. A vertex encoding unit(130) encodes at least one vertex which is removed through the simplification by using covariance matrix. An entropy encoding unit(140) converts the encoded base mesh and the encoded vertex into a bit stream by encoding the mesh and the vertex.
Abstract:
PURPOSE: A three-dimensional mesh encoding method and an apparatus thereof are provided to determine a priority of each gate and to simplify a three-dimensional mesh based on a gate of the determined high priority. CONSTITUTION: A mesh simplification unit(602) includes a priority determining unit and a vertex removing unit. The mesh simplification unit simplifies a three-dimensional mesh by using the priority of a gate. The gate constitutes the three-dimensional mesh. A mesh encoding unit encodes connectivity information of the removed vertex through simplification. The vertex configures the three-dimensional mesh. [Reference numerals] (601) Three-dimensional mesh; (602,603) Mesh simplifying unit; (604) Start gate determining unit; (605) Vertex removing unit; (606) Priority determining unit
Abstract:
PURPOSE: A three dimensional mesh variable encoding apparatus is provided to minimize errors between original data and encoded data by encoding vertex location information of a three dimensional mesh. CONSTITUTION: A three dimensional simplifying unit(101) simplifies three-dimensional mesh. The three dimensional mesh simplifying unit creates vertex location and connection information by levels. An importance calculation unit(102) calculates the significance of the vertex position information. A three dimensional mesh encoding unit(103) encodes the three dimensional mesh considering of the significance of the vertex location information.
Abstract:
A moving picture object boundary estimating method and an apparatus thereof are provided to be applied to the real time video communication as being suitable for a portable terminal with small calculation amount. A face recognizer(211) extracts the center coordinates of the face of an object and the size information of the face from a moving picture frame. A edge detector(212) extracts an edge of the moving picture frame. An object divider(213) estimates the boundary of the head of the object through a semicircle, and estimates the boundaries of a shoulder and arm of the object through the second order function graph.
Abstract:
PURPOSE: A data processing device and a method thereof are provided to efficiently compress location information of a vertex which configures a 3D object. CONSTITUTION: A vertex partitioning unit(120) partitions a plurality of vertexes into one or more groups. The vertexes are included in a 3D object. A prediction mode determining unit(130) determines a location prediction mode for compressing a vertex location for at least one group. An encoding unit(140) encodes an identification index of the determined prediction mode and a prediction error vector for at least one group.
Abstract:
A method for estimating an object boundary of a moving picture is provided to estimate a boundary of a background and a user object during a video call or video communication. In the first frame of a moving picture, an initial object boundary is presumed so that an object and a background are divided(210). Each color probability density function is estimated about the divided object and background(220). A boundary strip having a predetermined width is generated around an object boundary of the previous frame along with the object boundary of a previous frame in a current frame. A coherence probability density function of the boundary strip is presumed(230). Energy minimization is performed(240).
Abstract:
데이터처리장치가제공된다. 정점분할부는압축하고자하는 3D 객체에포함된복수개의정점을적어도하나의그룹으로분할하고, 예측모드결정부는상기적어도하나의그룹의각각에대해정점위치를압축하기위한위치예측모드를결정할수 있다. 그리고, 부호화부는상기적어도하나의그룹의각각에대해결정된예측모드의식별인덱스및 예측오차벡터를부호화할수 있다.