Title: 基於影像處理技術之煙霧偵測
Smoke Detection Based on Image Visual Features
Authors: 邱詩婷
Keywords: 煙霧偵測;煙霧特徵;電腦視覺;影像辨識;smoke detection;features;computer vision;pattern recognition
Issue Date: 2010
Abstract: 監控系統種類繁多,本論文著重於煙霧偵測。煙霧是個看得見,卻摸不清的東西,不論是運動方向或運動模式,都是無法明確的表達、闡述。本研究將以影像的資訊,探討煙霧的特徵。 本論文的主要架構為「煙霧特徵介紹」、「演算法介紹」以及「實驗結果分析」。在特徵介紹中,為了使系統能夠適應各種不同的環境,本論文著重於煙霧與背景的關聯性,探討煙霧的產生前後,對於畫面以及背景的影響,並設計各種規範,實現自動化煙霧偵測系統。 在於實驗結果分析,本研究設計各種實驗場景,由自然光的環境到光線不勻的隧道,由室外到室內,由白天到夜間,並與先前的相關研究比較分析,探討其相異之處。
Vision is one of the most direct receptors. The current surveillance that is still a human equipped with video monitors. Highly homogeneous work like this, the human beings will be out of attentions and become poor efficiency. This paper will introduce an automatic monitor system to complete the associated mission. It will save on the cost and also have related work efficiently without an exhausted condition. There are lots of kinds of monitoring systems. This paper focuses on smoke detection. In this paper consists of "smoke features introduction", "Introduction to smoke detection algorithm" and "experimental results". About introduction, in order to adapt to different environments, our work focuses on the correlations of smog and background. Then design the rules to implement smoke detection systems. Analysis of experimental results, we design various experimental scenes, the natural environment, Muja tunnel, outdoor, indoor, and nighttime. We will take a look of the previous related research and analysis, and discuss the differences.
Appears in Collections:Thesis