112AA074

112

14,766 千元

土壤及地下水污染調查、應變及整治之推動

2023/02/03

2023/11/30

王耀銘

環管署環境執法組

劉姵含

財團法人工業技術研究院

廢棄物管理

清運車輛軌跡分析、棄置風險分析、UAV 監控、車牌辨識

Truck trajectory analysis, Disposal risk analysis, UAV surveillance, License plate recognition

李唯祺

王耀銘

03-57332675#

AlexWang@itri.org.tw

配合環境管理署需求，本計畫執行包括非法棄置潛勢熱區分析平臺系統維運、場址空間關聯性分析及產出風險場域地圖、運用清運機具GPS 軌跡建立風險路徑分布特徵模型並產出非法棄置風險預警名單、以及應用遙控無人機及道路監視器等科技工具進行特定區域遠端監測管理等工作。本計畫利用人口密度、地表高程、道路、環保設施等環境因子與WDMS 場址之空間關係，建立特徵參數風險地圖，並驗證約77.85% WDMS 新增場址與67.94%的衛星影像變異點位於風險地圖的中度風險以上區域。上述風險地圖與其他四大查詢主題已建置於分析平臺系統，並以視覺化圖資提供直覺化操作與環域關聯效果。 本計畫運用110 年度約294 萬筆清運機具軌跡數據，分析行車軌跡路徑、停頓點及速度等特徵，建立路徑風險演算模型，最終篩選出241 筆異常軌跡分段預警名單，其中62 筆軌跡分段發現停等風險異常、啟動位置異常的現象。此外，本計畫運用無人機進行北區5 處、中區5 處、南區10 處場址的預警重點對象遠端監 測，產出正射影像鑲嵌圖、數值地表模型及場址三維量體模型。配合環境管理署推動非法棄置智慧圍籬系統平行監控，選定台61 線4 處交流道架設監控設備進行車牌辨識通報與GPS 軌跡比對，以智慧圍籬系統379 輛異常車輛進行關聯分析，發現14 輛極可疑異常名單及24 輛可疑關聯名單。

清運車輛軌跡分析、棄置風險分析、UAV 監控、車牌辨識

112AA074_成果報告.pdf

0MB

112AA074_成果報告公開版

2024/2/10 上午 12:00:00

For the requirements from EMA （Environmental Management Administration）,this project consists of several works, including the maintenance and operation of an illegal dumping potential zone platform, spatial correlation analysis and risk maps for dumping sites, creation of characteristic models for GPS trajectory of transport machinery and an illegal abandonment risk alert lists, as well as the applications of technological tools such as drones and monitors for remote monitoring and management. To achieve these goals, this project first employs the spatial relationship between WDMS （Waste Dumping Management System ） sites and several environmental factors such as population density, elevation, roads, and environmental facilities to establish several risk maps. This project identified approximately 77.85% of new WDMS sites, and 67.94% of new satellite variation points located in area of the above-moderate risk on the risk map. Additionally, to provide intuitive operations and environmental correlation effects, the project developed a platform that allows users to query all risk maps and related four topics with visual technologies. This project collected 2.94 million trajectory data to analyze driving trajectory paths, stop points and other characteristics, and establish path risk model for transport machinery. A total of 241 abnormal trajectory segments were screened out, of which 62 trajectories were found to have abnormal stops and abnormal starting positions. Furthermore, drones were used to monitor 5 sites in the Northern, 5 sites in the Central and 10 sites in the Southern regions, and produce orthophoto mosaics, elevation models, and three-dimensional volumes for these sites. In cooperation with the EMA's promotion of parallel monitoring of illegally abandoned smart fence system, this project installed monitoring equipment at 4 intersections of Provincial Line 61 for license plate recognition and GPS track comparison. The correlation results of 379 abnormal vehicles from smart fence system were found a list of 14 extremely suspicious vehicles and 24 suspiciously related vehicles.

Truck trajectory analysis, Disposal risk analysis, UAV surveillance, License plate recognition