摘要
图像火检算法作为火灾预警与监测的关键技术,在复杂环境下的性能尤为重要。文章针对图像火检算法在复杂环境下的性能进行评估,发现强光、火焰形状变化影响其准确率。为改善此问题,提出预处理优化、特征精确提取及算法融合与集成学习三大策略。实施后,算法在复杂条件下的准确率得到显著提升,验证了优化策略的有效性,为火检算法性能评估提供了系统方法,优化策略对提升火灾预警与监测的准确性和鲁棒性具有重要价值。
The image fire detection algorithm,as a key technology for fire warning and monitoring,has particularly important performance in complex environments.This study evaluates the performance of image fire detection algorithms in complex environments and finds that strong light and changes in flame shape affect their accuracy.To improve this problem,three strategies are proposed:preprocessing optimization,precise feature extraction,algorithm fusion,and ensemble learning.After implementation,the accuracy of the algorithm was significantly improved under complex conditions,verifying the effectiveness of the optimization strategy.This study not only provides a systematic method for evaluating the performance of fire detection algorithms,but its optimization strategy has important value in improving the accuracy and robustness of fire warning and monitoring.
作者
叶红武
张林
张治湖
宋继坤
YE Hongwu;ZHANG Lin;ZHANG Zhihu;SONG Jikun(Guoneng Changyuan Hanchuan Power Generation Co.,Ltd.,Wuhan 431614,China;Yantai Longyuan Power Group Co.,Ltd.,Yantai 265503,China)
关键词
图像火检算法
性能评估
预处理技术
image flame detection algorithm
performance evaluation
preprocessing techniques
