Abstract:
PURPOSE: A method for detecting the movement of a face based on an optical flow and a recording medium thereof is provided to analyze the movement of the face in real time. CONSTITUTION: A method for detecting a movement of a face based on an optical flow comprises steps of dividing a face area and matching the face area to a three-dimensional mesh model, and changing the three-dimensional face model using an extracted motion parameter. In the method, the process for dividing the face area comprises steps of deciding a reference map by learning, dividing an image using the reference map, extracting a feature to remove a noise, and dividing the face area. The matching process comprises steps of extracting a feature point from an image, matching the feature point to a generic mesh model, and moving a node of the generic mesh model using a linear interpolation to perform a matching process of each portion. The optical flow is calculated by a differential way.
Abstract:
PURPOSE: A mixed mode delay locked loop with quick synchronizing time and small jitter is provided to enable users to apply the device to chips which require high speed data transmission or low electric power. CONSTITUTION: A mixed mode delay locked loop with quick synchronizing time and small jitter includes a Voltage Controlled Delay Line(VCDL) of an analog circuit unit and a Fixed Delay Line(FDL) of a digital circuit unit. An analog circuit unit also includes a Phase Frequency Detector(PFD), an Internal Clock Detector(ID), a Charge Pump(CP), a 1/2Vcc generator, a loop filter, a Voltage to Current Converter, and a Voltage Controlled Delay Line(VCDL). The FDL of digital circuit unit includes a Replica of a clock driver which is a Monitor Driver, a Time to Digital Converter, and a Digital to Time Converter.
Abstract:
PURPOSE: A facial expression method is provided to express more natural and realistic face look and automate a facial expression animation by using a spring mass damper model based on a physical property and a muscle model based on a parameter base model. CONSTITUTION: A facial expression method comprises steps of generating a polygonal face mesh model with a plurality of nodes and edges, assuming each edge as a spring by applying a spring-mass-damper model to a facial mesh model, obtaining a dynamic equation modeling a force transferred to an adjacent node via the edge, imposing a force to an arbitrary part of the facial mesh model according to a facial expression, and limiting a force transfer area when a force is transferred to an arbitrary part of the facial mesh model. If the force is imposed to the arbitrary part of the mesh model, the force is transferred to a limited area of the adjacent node according to the dynamic equation.