Online Visual Tracking by Huchuan Lu & Dong Wang

Author:Huchuan Lu & Dong Wang
Language: eng
Format: epub
ISBN: 9789811304699
Publisher: Springer Singapore

First, image representation contains two aspects, that is, features (e.g., color, texture, and Haar-like feature) and the fashion of using them (e.g., histogram). Our image representation consists of a set of linearized RGB color space and their histogram representation.

Second, the appearance model is composed of a discriminative classifier, the output of which is a probability () rather than a binary value, where x represents an individual pixel and y is a binary value indicating the pixel belonging to the object of interest () or its surrounding background (). Using the discriminative classifier at a given frame, every pixel is mapped to a probability value, and a probability map is then built. The peak of the probability map convolved with an Epanechnikov kernel [5], which is related to the mean shift kernel, is selected as the center of the moving object.

Fig. 5.2Motion model in our tracking system

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.