|
Face Detection and its Applications
|
| Facial feature detection is a hot research area in both computer vision and computer graphics. Inspired by LFA (Local Feature Analysis) and AAM(Active Appearance Model), we present here a novel facial feature detection algorithm based on hierarchical constraints. Results demonstrate the algorithm is efficient and robust. The algorithm first locates the approximate face position using Aaboost; then uses a coarse global model to find initial position of different facial features; the algorithm finally searches the precise features using delicate local feature model. By automatically searching corresponding points, we greatly reduce tedious interaction in face morphing and get quite nice results. |
| Examples: |
|
|
|
Facial features location. |
| Face morphing is one of the most widely used graphics techniques in ˉlm making, game design, video conference, etc. Smooth morphing relies on the accurate feature correspondence between source image and destination image, otherwise, it will cause so-called \ghost effects". Tradition methods need tedious user interaction to locate the facial features. In this paper, we propose a feature detection based automatic face morphing framework without any need of user interaction. The algorithm first locates the approximate face position using Aaboost; then uses a coarse global model to find initial position of different facial features; the algorithm finally searches the precise features using delicate local feature model. By automatically searching corresponding points, we greatly reduce tedious interaction in face morphing and get satisfactory results. |
| Examples: |
 |
|
Automatic morphing using face detection. |
| This page is a cooperative project with Feng Tang who is now pursuring Ph.D in UCSC. |
|
|
