嵌入式攝影機系統之白天與夜晚火焰偵測技術

Daytime and Nighttime Flame Detection Technology in Embedded Camera System

張嘉育(Jia-Yu Jhang)

指導教授 ： 廖珗洲

朝陽科技大學/資訊學院/資訊工程系/碩士(2015年)

摘要

隨著科技的進步，攝影機、監視系統的應用在日常生活中愈來愈廣，嵌入式系統硬體設備、效能也不斷的提升，使得愈來愈多功能可以直接在嵌入式攝影機系統上實現，而不需要其他的硬體設備。
智慧型監控是監視系統一個重要的發展方向，因此鼎高科技與朝陽科大資工系進行智慧型影像監控方面的系統開發，期望透過產學合作案提升其產品的附加價值。因此本研究進行嵌入式攝影機系統的火焰偵測技術開發。依據攝影機可能擷取到的圖片色彩，將之分為白天彩色影像及夜間灰階影像兩大類，分別設計不同偵測流程。
白天彩色影像偵測流程中主要進行顏色過濾以及閃爍偵測來判斷影是否有火焰存在。而針對夜間灰階影像則是進行高亮度偵測、移動物體偵測及閃爍偵測三個偵測步驟。透過現有攝影機上的亮度感測器可以直接決定採用白天或是夜晚的火焰偵測方法。
為了讓偵測方法可以融入實際監視系統中，我們將火焰偵測的靈敏程度分為8個等級；攝影機的拍攝場景分為遠、中、近。針對8個等級、3個場景設計不同的偵測參數，降低使用者參數設定上的困擾。
實驗結果顯示，白天火焰偵測的偵測時間約為7~20秒。夜間火焰偵測的偵測時間約為5~7秒。

關鍵字

小波轉換 ； 顏色偵測 ； 火焰偵測 ； 嵌入式系統 ； 影像處理 ； 移動物體偵測

並列摘要

The cameras and video surveillance systems are getting popular and ubiquitous. The computing power of the embedded camera system is also increasing with the growth of hardware technology. It enables many image-based functions can be realized in the embedded camera system without any additional hardware, such as personal computer.
Intelligent video surveillance (IVS) is an important development direction. An industrial-academic project was funded by Telexper Co. Ltd. to generate the value-added function of IVS in an embedded camera system. In this study, a flame detection technology was developed to detect flame at the daytime or nighttime. Two processes are designed for the flame in color at the daytime and flame in grayscale at the nighttime.
The process of the flame detection at the daytime consists of two steps: color filtering and flicker detection. The process of the flame detection at the nighttime consists of three steps: bright area detection, motion detection, and blinking detection. The built-in ambient light sensor of the camera system can be used to determining the execution of the daytime or nighttime flame detection process.
Besides, eight sensitivities and three kinds of monitoring areas are designed. All the parameter settings for every sensitivity and monitoring area are also designed to increase the feasibility of the flame detection technology.
The experiment results show that the detect time of daytime is about 7~20 seconds and detect time of nighttime is about 5~7 seconds.

並列關鍵字

Wavelet Transform ； Color Detection ； Flame Detection ； Embedded System ； Image Processing ； Moving Object Detection

參考文獻

[2]B. U. Töreyin, Y. Dedeoğlu, U. Güdükbay, and A. E. Çetin, "Computer vision based method for real-time fire and flame detection," Pattern Recognition Letters, vol. 27, 2006, pp. 49-58.

[3]J. Chen, Y. He, and J. Wang, "Multi-feature fusion based fast video flame detection," Building and Environment, vol. 45, 2010, pp. 1113-1122.

[5]T. Xuan Truong and J.-M. Kim, "Fire flame detection in video sequences using multi-stage pattern recognition techniques," Engineering Applications of Artificial Intelligence, vol. 25, 2012, pp. 1365-1372.

[7]I. Kolesov, P. Karasev, A. Tannenbaum, and E. Haber, "Fire and smoke detection in video with optimal mass transport based optical flow and neural networks," in Image Processing (ICIP), 2010 17th IEEE International Conference on, 2010, pp. 761-764.

[8]A. Stadler, T. Windisch, and K. Diepold, "Comparison of intensity flickering features for video based flame detection algorithms," Fire Safety Journal, vol. 66, 2014, pp. 1-7.