| Summary: | 碩士 === 國立清華大學 === 資訊工程學系 === 93 === This thesis presents an example-based method to
synthesize a human portrait sketch from the given photo. There
are two major modules in the proposed method：the feature
extraction module, and the synthesis module. In the feature
extraction module, we take advantage of the trained Active
Appearance Model (AAM) to automatically extract feature
points from the given example. Then we compute feature
vectors of face components, such as eyes, nose, lip, and so on
according to the geometric positions of feature points. In the
synthesis module, we first compute the feature vectors of the
given photo. Then, based on these feature vectors we find the
most similar drawing component of each face component in the
example database. With these retrieved face components and
user selected hair style, we construct the final human facial
sketch according to the placement parameters obtained from
feature vectors. Several experimental examples are given to
demonstrate the effectiveness of the proposed method.
|
|---|
