Video Tracker Models
Abstract
This preliminary report considers modelling of two classes of video trackers from their algorithmic and measurement aspects. Because they are widely used and easily implementable in hardware, the classes chosen were the binary centroid tracker and the binary correlation tracker. Many variations exist even in these classes; their value in the present setting is to demonstrate the modelling aspects and the salient features that may be derived from an analysis of the behavior of these systems in the presence of noise. The models predict tracker offset errors (mean error) and track point jitter (the standard deviation of the error). Track performance, in general, is dependent on the character of the object pattern being tracked and on the distribution of the noise associated in the formation of the array of pixels which represent the pattern. In the analyses which follow, a simple square pattern is used and Gaussian additive noise is assumed. The models permit any geometric pattern as input and the question of noise distribution and false alarm in the formation of the binary pattern from the linear data input. Additional keywords: Computations, and Equations.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 1983
- Accession Number
- ADA151733
Entities
Organizations
- Westinghouse Electric Corporation