一种基于形状的红外图像泄漏气体检测方法

Shape-based infrared image leakage gas detection method

  • 摘要: 针对工业生产中泄漏气体导致的爆炸和火灾问题, 提出一种基于形状和SVM分类的红外图像泄漏气体检测方法。采用泄漏气体和干扰物红外图像样本的形状特征训练SVM分类器, 通过对红外图像序列采用基于背景差分的运动检测得到候选目标区域, 再对候选目标区域提取其形状特征, 最后使用SVM分类器进行判别, 从而得到最终的检测结果。使用乙烯气体泄漏仿真数据进行实验, 检测率最高可达98%, 结果表明, 采用该方法可以有效检测泄漏气体, 相比其他方法, 极大地减少了干扰物造成的误检。

     

    Abstract: Aiming at the explosion and fire caused by leakage gas in industrial production, an infrared image leakage gas detection method based on shape and support vector machine(SVM) is proposed. The SVM classifier is trained by using the shape features of the infrared image sample of the leaking gas and the interfering object. The candidate target region is obtained by using the background difference-based motion detection for the infrared image sequence, and then the shape feature is extracted from the candidate target region, and finally the SVM classifier is used to obtain the final detection result. Experiments were carried out using ethylene gas leakage simulation data, and the detection rate was up to 98%. The results show that this method can effectively detect the leakage gas, which greatly reduces the false detection caused by the interference.

     

/

返回文章
返回