IMAGE SEQUENCES TEXTURE ANALYSIS BASED ON VECTOR QUANTIZATION
Sergii Bogucharskiy, Sergii Mashtalir · Radio Electronics Computer Science Control · 2014
The approach for the image sequence (videodata) analysis is proposed. For this purpose, the matrix analogs of existing neural network approaches is developed. This aallows to takes into account the spatial relationships of multimedia information, and to reduce the time of information processing through the introduction of a new neural network matrix training procedures. The texture isselected as a basic characteristic for original data partition, which, in turn, improves the clustering accuracy.