One of the well known risks of large margin training methods, such as boosting and support vector machines (SVMs), is their sensitivity to outliers. These risks are normally mitig...
—In this paper, we present a design architecture of implementing a ”Vector Bank” into video encoder system, namely, an H.264 encoder, in order to detect and analyze the movin...
— the paper discusses an approach of using traditional time series analysis, as domain knowledge, to help the data-preparation of support vector machine for classifying documents...
Ting Yu, Tony Jan, John K. Debenham, Simeon J. Sim...
In this contribution, we present an algorithm for lowcomplexity global motion estimation, that works with block-coded video (e.g. MPEG-2). A superimposed global motion model is fi...
Snakes, or active contours, are used extensively in computer vision and image processing applications, particularly to locate object boundaries. Problems associated with initializ...