EXPLORATORY STUDY ON INTELLIGENT MULTI-RADAR MULTI-TARGET TRACKING

PDF

Authors
  1. Wong, A.K.C.
  2. Lin, H.
Corporate Authors
Waterloo Univ, Waterloo ONT (CAN) Pattern Analysis and Machine Intelligence Lab;Defence Research Establishment Ottawa, Ottawa ONT (CAN)
Abstract
Describes a target association methodology known as Constellation Matching (CM) and its application to multiple target tracking. The CM method is different from the traditional data association methods such as NNSF, MHT and JPDA in the sense that it makes full use of the target spatial configuration information for the plot-to-target assignment. To speed up the CM process, two heuristics are incorporated: 1) the use of spatial constraints to eliminate impossible assignments and 2) the use of maximum entropy partitioning method to subdivide the graph representation into subgraphs with lower order to increase the matching speed and reduce the computation complexity. The CM method is then combined with Kalman Filter techniques for target tracking as well as further data association. The efficiency and accuracy of one depends on those of the other. To demonstrate and evaluate the capability of the CM-based MTT system developed, real world radar data provided by DREO are used. The experimental results show that the CM-based MTT system is effective in accomplishing its task in most cases even in noisy environment. For comparison purpose, an MTT system based on Multiple Hypothesis Testing (MHT) is also implemented in this report. The results on the same data set are obtained using both the MHT-based MTT system and the CM-based MTT system. The comparison show that in most cases, CM-based MTT system yields better results.
Keywords
CONSTELLATION MATCHING (CM);MULTIPLE TARGET TRACKING;MULTIPLE HYPOTHESIS TESTING
Report Number
DREO-CR-94-641 — Final Report
Date of publication
01 May 1994
Number of Pages
42
DSTKIM No
95-01188
CANDIS No
148561
Format(s):
Hardcopy;Document Image stored on Optical Disk

Permanent link

Document 1 of 1

Date modified: