Video Segmentation Based On Descriptors Extracted from Wavelet Transforms

• Grant number: 09/10627-5
• Support Opportunities: Scholarships in Brazil - Doctorate
• Start date: January 01, 2010
• End date: September 30, 2013
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computational Mathematics
• Principal Investigator: Hélio Pedrini
• Grantee: Ricardo Dutra da Silva
• Host Institution: Instituto de Computação (IC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Associated scholarship(s): 12/01002-4 - Image analysis using morse theory and homological persistence, BE.EP.DR

Abstract

Video segmentation is a fundamental process in several applications such as automatic video retrieval, robotic vision, industrial surveillance and forensic science. The main objective is to allow a meaningful discrimination among interest objects. Traditional techniques to extract texture based on pixels usually demand high computational cost. This work presents a segmentation method based on wavelet transforms to extract spatial luminance and chrominance information from videos. To reduce computational cost, the method comprises two stages. Initially, each image is decomposed in blocks and a wavelet transform is applied to each of these blocks to identify homogeneous regions of the image, assigning the whole block to its respective class. A refinement stage is applied in the remaining pixels, detected at the first stage as belonging to heterogeneous regions. Since the method operates without training sets, besides reducing costs, the first stage also extracts descriptors for each object. The wavelet transform is intended to capture objects movement and transitions between video shots. Preliminary studies are developed to evaluate the method with quantitative metrics and comparisons with other existing methods.

Academic Publications

(References retrieved automatically from State of São Paulo Research Institutions)

SILVA, Ricardo Dutra da. Complexo discreto de Morse para imagens: algoritmos, modelagem e aplicações. 2013. Doctoral Thesis - Universidade Estadual de Campinas (UNICAMP). Instituto de Computação Campinas, SP.