基于火焰状氧化铝回转窑火焰图像检索

Alumina rotary kiln flame image retrieval based on flame shapes

提出一种基于火焰形状的氧化铝回转窑火焰图像的检索方法,对回转窑火焰的形状进行深入分析,根据其特点,提出一种描述火焰形状的基准距离函数,通过傅里叶变换得到火焰图像形状的傅里叶描述子。结果表明:该距离函数能较好地描述火焰形状,应用此火焰形状描述子,设计一套火焰图像检索原型机系统。以现场采集的火焰图像作为实验对象,将本研究提出的方法与传统方法和不变矩方法的进行检索实验。结果表明:本研究提出的方法具有较高的检索精度,且明显优于传统方法和不变矩方法。本研究的结果可为回转窑生产控制提供有力支持。

A shape based alumina rotary kiln flame image retrieval method was proposed. First, the flame shapes were analyzed detailedly, a distance function was put forward. Through Fourier transformation, the flame shape Fourier descriptors were calculated. The results indicate that this distance function can depict the shape effectively. A flame image retrieval prototype system was proposed using this Fourier descriptors. With the flame images sampled from an alumina rotary kiln, retrieval experiments were carried out with the traditional retrieval method, invariant moment retrieval method and the proposed method. The results demonstrate that the proposed method outperforms the other two methods in terms of retrieval precision and efficiency. The research results can provide strong support for the control of the rotary kiln.