Field and Mesh-Based Automatic Morphing of Face Images Using Feature Points Extraction
Yong‐Ki Kim, Byeong-Rae Lee · Jeongbo gwahaghoe nonmunji. so'peuteuweeo mich eung'yong · 2011
This paper presents an efficient and automatic method dealing with morphing of frontal color facial images, based on a combination of field morphing and mesh morphing. The proposed field and mesh-based morphing is automatically conducted by using feature points of facial images. By analyzing certain features of facial skin color, we can determine a specific range of skin colors in a normalized histogram. After detecting the eye candidates based on eye-template matching, respective candidate areas are calculated in proportion of the distance between pupils. For respective candidate areas, the target facial feature points are extracted by making use of edge maps, grey level gradients and analyzing histograms, etc. Control lines are made up of the extracted feature points. Source mesh control points are made in the source images at regular intervals. The field and mesh-based morphing is automatically conducted, calculating the target mesh control points for the source mesh control points using field warping algorithm. The proposed field and mesh-based morphing has the advantage of computational time that is 80% faster than that of the automatic field-based morphing. In addition, it has the similar quality of the intermediate frames from the field morphing and has removed the manual movemnet of source mesh control points into its target.