Research Article

Collaborative Research on Mouth Shape and Lyrics in Singing Practice Based on Image Processing

Table 2

Comparison of three extraction methods.

MethodTypical algorithmAdvantagesDisadvantages

Shape-based methodGeometric feature method, snake model method, and active contour model methodLow feature dimension; it is not easy to changeThe lip movement information reflected is not comprehensive; the image has clear edges
Pixel-based methodMethod based on image transformationAll pixels represent visual information together, and the loss of information is relatively smallHigh feature dimension; image geometric transformation sensitivity
Hybrid methodsMethod based on motion analysisCombine shape-based and pixel-based featuresThe algorithm is complex; it is difficult to extract the ideal contour in processing