The bounded variation quantity (B.V.Q.) and its application to feature extraction

作者:

Highlights:

摘要

The concept of bounded variation quantity (B.V.Q.) is proposed as a local feature of patterns, especially waveform patterns and picture patterns. It is closely connected with the mathematically complicated concepts of total variation and distribution, but the result is shown by a simplified form (i.e. a summation of absolute differences) for concrete calculations. Its availability is also demonstrated through several applications such as seismic-wave analysis, the estimation of standard deviation in an independent Gaussian process and a method of detecting the directionality of picture patterns.

论文关键词:Bounded variation quantity (B.V.Q.),Local feature,Theory of Distribution,Seismic-wave analysis,Real-time processing,Gaussian noise,Convergence ratio,Directionality,Texture analysis,Probabilistic set theory

论文评审过程:Received 13 May 1980, Revised 22 December 1980, Available online 19 May 2003.

论文官网地址:https://doi.org/10.1016/0031-3203(82)90004-8