水位影像量測技術視窗化之研究

Windowing of the Water Level Image Measuring Techniques

王傳益(Chuan-Yi Wang) ； 楊翰宗(Han-Chung Yang) ； 何璟浩(Ching-Hao Ho) ； 林長葵(Chang-Kuei Lin)

《農業工程學報》 62卷2期 (2016/06) Pp. 54-63

https://doi.org/10.29974/JTAE.201606_62(2).0005

摘要

本研究目的在於探討水位影像量測技術視窗介面化，藉由高解析度相機進行水位影像拍攝，並選用Matlab 作為水位影像處理技術之開發程式語言，利用其影像處理與辨識技術優勢辨識水位；其後結合Visual Basic 所撰寫成的視窗化介面程式整合成多功能軟體，並藉由滑鼠點擊影像辨識之水位線得出即時之水位。相較於傳統之水位計具有價廉及影像可視化之優勢。研究結果顯示，於模擬夜間降雨且為動態水流情況下，其水位平均絕對誤差百分比(MAPE)均小於1.50%。復於新店溪中游處之粗坑壩進行現地試驗，將視窗化介面程式判讀所得水位值與當地壓力式水位計進行比較，結果顯示平均絕對誤差百分比(MAPE)皆小於1.00%。整體而言，影像辨識視窗化介面技術應用於現地即時動態水位量測應具可行性，且其量測所得水位資料具可靠度。

關鍵字

視窗化 ； 影像處理

並列摘要

The objective of this study is to investigate the water level image measuring techniques with windowing interface by using high-resolution camera. Matlab programming language was selected for image processing and recognition. Identification code of water level combines windowing of the multifunctional integrated interface software with Visual Basic. The current water level can be obtained by clicking the mouse at the water level line. The image processing and recognition technologies provide visualization of real-time water level, which is cheaper than the traditional water gauge. To consider a severe environmental condition with a rainfall under dynamic flow during the nighttime, the mean absolute percentage errors (MAPE) are less than 1.50%. Furthermore, we analyzed the image recognition of water level with field tests at Tsu-Keng dam in the middle reaches of the Xin-Dian Creek. The mean absolute percentage error (MAPE) of water level are less than 1.00% between the values by the image recognition system and those by the local pressure water gauge. Overall, windowing interface image recognition system with real-time dynamic technology is feasible and reliable for field water level measurements.

並列關鍵字

Windowing ； Matlab ； Image processing ； Visual Basic