Geometry compression of 3-D mesh models using predictive two-stage quantization

Jin Soo Choi, Yong Han Kim, Ho Jang Lee, In Sung Park, Myoung Ho Lee, Chieteuk Ahn

Research output: Contribution to journalArticlepeer-review

15 Scopus citations


In conventional predictive quantization schemes for 3-D mesh geometry, excessively large residuals or prediction errors, although occasional, lead to visually unacceptable geometric distortion. This is due to the fact that they cannot limit the maximum quantization error within a given bound. In order to completely eliminate the visually unacceptable distortion caused by large residuals, we propose a predictive two-stage quantization scheme. This scheme is very similar to the conventional DPCM, except that the embedded quantizer is replaced by a series of two quantizers. Each quantizer output is further compressed by an arithmetic code. When applied to typical 3-D mesh models, the scheme performs much better than the conventional predictive quantization methods and, depending upon input models, even than the MPEG-4 compression method for 3-D mesh geometry both in rate-distortion sense and in subjective viewing.

Original languageEnglish
Pages (from-to)312-322
Number of pages11
JournalIEEE Transactions on Circuits and Systems for Video Technology
Issue number2
StatePublished - 2000


Dive into the research topics of 'Geometry compression of 3-D mesh models using predictive two-stage quantization'. Together they form a unique fingerprint.

Cite this