多感測器運動／影像模糊資料相關之融合演算

The Fusion Algorithm of Multisensor Kinematic/Image Fuzzy Data Association

陳以明(Y. M. Chen) ； 黃煌哲(H. C Huang)

《中國航空太空學會學刊》 32卷3期 (2000/09) Pp. 243-250

https://doi.org/10.6124/TAASRC.200009_32(3).06

摘要

在多感測器資料相關之融合演算，傳統的非貝氏法或貝氏法之資料相關法受限於僅能以感測器之運動型態的量測訊息（如距離、方位）進行資料融合演算，對於感測器其它非運動型態的訊息則無法一併作資料相關演算。本文以雷達／紅外線感測器爲例，提出一個以模糊邏輯爲基礎稱之模糊資料相關法，可將感測器量測之運動型態與非運動型態（彫像）的訊息一併作資料相關演算。同時爲了簡化融合演算的複雜性，將模糊資料相關法劃分爲模糊運動型態資料相關與模糊影像型態資料相關，分別計算其相關機率值再以權重的方式予以整合。模擬結果驗證模糊資料相關法優於傳統以貝氏法爲基礎的聯合機率資料相關法；此外，增加非運動型態的影像資料融合估算結果能有效提昇追蹤性能。

關鍵字

模糊邏輯 ； 資料相關 ； 目標追蹤

並列摘要

There is a limitation to process data fusion by means of traditional non-Bayesian or Bayesian association algorithm of multisensor data fusion (MSDF). It's only kinematic measurement data (such as range, azimuth) used in processing, however, MSDF doesn't use the other useful data-non-kinematic measurement data, This paper proposes a fuzzy data association (FDA) algorithm based on fuzzy logic for Radar/Infrared sensor data fusion. The FDA combine non-kinematic (Image) data with kinematic measurement data. We decompose FDA into fuzzy kinematic data association (FKDA) and fuzzy image data association (FIDA) in order to simplify the fusion algorithm. We calculate association probabilities of FKDA and FIDA individually and then integrate those probabilities by weighting method. Results of test show that performance of FDA is superior to JPDA based on a Bayesian approach. Moreover add non-kinematic data in FDA can improve tracker performance effectively.

並列關鍵字

fuzzy logic ； data association ； target tracking