以影像的最大能量位移進行火災偵測

Fire Detection with Maximum Energy Displacement of Images

江佳翰(CHIA-HAN CHIANG)

指導教授 ： 涂世雄

中原大學/電機資訊學院/電機工程研究所/碩士(2021年)

https://doi.org/10.6840/cycu202100512

摘要

在本文中，提出一種基於影像的最大能量位移進行火災偵測的方法。以攝像頭取得影像，進行影像處理，加入最大能量位移，將其變化量作分析，進行火焰辨識。
本論文分為四個部分，第一部分我們使用攝像頭獲取的影像，進行圖像處理來獲得每張圖像各自的R、G、B值。然後利用Python求出每張色彩圖像的最大能量位移區間，以此來進行火災辨識。第二部分，我們對RGB通道用OpenCV轉換成HSV通道，獲取HSV的最大能量位移區間，以HSV進行火災辨識能對色相與飽和度得到較為詳細的結果。第三部分，我們對RGB通道用OpenCV轉換成灰階通道，獲取灰階的最大能量位移區間，由於灰階相較於RGB只有單個通道，以灰階進行火災辨識能在不降低太多辨識率的同時提高辨識速度。第四部份，我們使用這個系統來檢測不同類型的圖像並進行分析，都得到了預期的實驗結果。
本文的研究貢獻如下:
(1)低成本:在本論文中，利用Python進行系統設計，降低了開發環境的額外花費。
(2)增加準確性:在本論文中，利用RGB、灰階、HSV多方面的最大能量位移同時進行相互比較，能提高偵測的準確度降低誤辨率。
(3)適應性:在本論文中，我們考慮了不同時間下的火焰狀況，讓系統能對黑夜與白天都能進行準確的影像辨識。

關鍵字

影像處理 ； 能量位移 ； 火災偵測

並列摘要

In this thesis, a fire detection method based on the maximum energy displacement of the image is proposed. We use the camera to obtain the image, perform image processing add the maximum energy displacement, analyze the change amount and perform flame identification. This thesis is divided into four parts. In the first part, we use the images acquired by the camera to perform image processing to obtain the respective R, G, and B values of each image. Then use Python to find the maximum energy displacement interval of each color image for the fire identification. In the second part, we use OpenCV to convert the RGB channels into HSV channels to obtain the maximum energy displacement interval of the HSV, use HSV for fire identification can produce more detailed results on hue and saturation. In the third part, we use OpenCV to convert the RGB channels into Grayscale channels to obtain the maximum energy displacement interval of the Grayscale. Since grayscale only has a single channel compared to RGB, using grayscale for fire identification can increase the identification speed without reducing the identification rate too much. In the fourth part, we use this system to detect and analyze different types of images, and each one gets the expected experimental results.

In this thesis,we have some contributions as follows:
(1)Low cost: In this thesis, we use Python to program and reduce the extraneous expense of the development environment.
(2)Increased accuracy: In this thesis ,we use RGB, grayscale, and HSV to detect and compare each other at the same time, which can improve the accuracy of detection and reduce the error rate.
(3)Adaptability: In this thesis, we consider the fire conditions in different cases, so that the system can accurately recognize images at night and during the day.

並列關鍵字

Fire Detection ； Energy Displacement ； Image processing

參考文獻

References

[1]唐唯銘，“以影像偵測執行及時火災偵測”，中原大學通訊碩士所，2020年6月。

[2]K. Chen, Y. Cheng, H. Bai, C. Mou and Y. Zhang, "Research on Image Fire Detection Based on Support Vector Machine," 2019 9th International Conference on Fire Science and Fire Protection Engineering (ICFSFPE), 2019, pp. 1-7.

[3]S. Mohd Razmi, N. Saad and V. S. Asirvadam, "Vision-based flame detection: Motion detection fire analysis," 2010 IEEE Student Conference on Research and Development (SCOReD), 2010, pp. 187-191.

[4]O.Giandi and R. Sarno, "Prototype of fire symptom detection system," 2018 International Conference on Information and Communications Technology (ICOIACT), 2018, pp. 489-494.