Structural analysis of videos with hidden markov models and dynamic programming

   
   

A method analyzes a high-level syntax and structure of a continuous compressed video according to a plurality of states. First, a set of hidden Markov models for each of the states is trained with a training video segmented into known states. Then, a set of domain specific features are extracted from a fixed-length sliding window of the continuous compressed video, and a set of maximum likelihoods is determined for each set of domain specific features using the sets of trained hidden Markov models. Finally, dynamic programming is applied to each set of maximum likelihoods to determine a specific state for each fixed-length sliding window of frames of the compressed video.

Un método analiza un sintaxis y una estructura de alto nivel de un vídeo comprimido continuo según una pluralidad de estados. Primero, un sistema de los modelos ocultados de Markov para cada uno de los estados se entrena con un vídeo del entrenamiento dividido en segmentos en estados sabidos. Entonces, un sistema de características específicas del dominio se extrae de una ventana que resbala de longitud fija del vídeo comprimido continuo, y un sistema de likelihoods máximos se determina para cada sistema de características específicas del dominio usando los sistemas de los modelos ocultados entrenados de Markov. Finalmente, la programación dinámica se aplica a cada sistema de likelihoods máximos para determinar un estado específico para cada ventana que resbala de longitud fija de bastidores del vídeo comprimido.

 
Web www.patentalert.com

< Channel tuning apparatus

< Method for capturing a panoramic image by means of an image sensor rectangular in shape

> Communicating information from an imaging device to a processor-based system

> Method apparatus and system for compressing data that wavelet decomposes by color plane and then divides by magnitude range non-dc terms between a scalar quantizer and a vector quantizer

~ 00166