B. Clip Synthesis
SSL considers only one feature that is hausdorf edge
feature for performing cartoon clip synthesis, hence the
clip synthesized by using SSL is not smooth. In case of
RCCS it uses two features, edge and motion direction
feature. The clips synthesized by semi-MSL is smoother
than SSL and RCCS, because RCCS neglects the color
information in the distance metric learning, the characters
in the synthesized results may have different colors and
SSL uses only edge feature, hence it is not efficient.
Hence in this section, we show that semi-MSL
outperforms the SSL used in GCCS and the UBDML used
in RCCS for cartoon clip synthesis.
B. Clip SynthesisSSL considers only one feature that is hausdorf edgefeature for performing cartoon clip synthesis, hence theclip synthesized by using SSL is not smooth. In case ofRCCS it uses two features, edge and motion directionfeature. The clips synthesized by semi-MSL is smootherthan SSL and RCCS, because RCCS neglects the colorinformation in the distance metric learning, the charactersin the synthesized results may have different colors andSSL uses only edge feature, hence it is not efficient.Hence in this section, we show that semi-MSLoutperforms the SSL used in GCCS and the UBDML usedin RCCS for cartoon clip synthesis.
การแปล กรุณารอสักครู่..
