Fire Flame Image Segmentation and Its Geometry Comparative Analysis
Fire detection technology, based on image processing, is one of important methods of early fire detection. Accurate extraction of flame images and selecting effective characteristics of target, which are very important to improve the accuracy of image-based early fire detection and reduce the occurrences of false positives or omissions. Related routines are wrote and run in matlab R2009b to process images collected. The process, which can get relatively integral segmental images from flame images obtained and has good robustness, has been determined after numerous experiments with several image segmentation methods. Five geometric features on compactness and complexity are calculated, including a series of early fire flame image segmentation and segmented images of disturbance. By comparison, the quite different geometric eigenvalues between Flames segmented images and interferential segmented images are eccentricity, solidity and rectangularity. Key characteristics selection can be provided basis by the conclusion on image type early fire flame detection.
Early fire Detection Flame Image Recognition Matlab
DING Xiebin HOU Zunze
Teams of Graduate Students, Chinese Peoples Armed Police Force Academy, Langfang 065000, Hebei, Chi Institute of the Chinese Peoples Armed Police Force Academy, Langfang 065000, Hebei, China
国际会议
The 2010 International Symposium on Safety Science and Technology(2010 安全科学与技术国际会议)
杭州
英文
835-840
2010-10-26（万方平台首次上网日期，不代表论文的发表时间）