基于CANNY算子的自适应双阈值油罐油位红外成像检测

Detection of IR imaging of auto-adaptive double-threshold tanks’ level based on Canny operator

针对油罐油位的监测问题,在研究CANNY边缘检测算子的基础上,提出了一种新的基于CANNY算子的自适应双阈值边缘检测方法。该方法能够根据不同的油罐油位红外成像图像的灰度梯度直方图的特征进行自动设置高低阈值,因此避免了传统的CANNY检测算子手工选择阈值不合理对边缘信息,特别是油位信息的丢失。在油位信息提取过程中,该方法能够自动分析横向条纹数目,并根据该数目获得每一条横向条纹的具体位置,使得边缘检测算法在油罐油位监测应用中更加实用。用该方法对一批油罐的红外成像图像进行了处理,结果表明所得油位的误差在±0.02m以内,而测量误差百分比为0.062%。

Aiming at the issue for the detection of tanks＇ liquid level, on the basis of analyzing the Canny edge detection operator, a novel auto-adaptive double-threshold edge detection method based on Canny operator is proposed. According to IR imaging gray gradient histogram＇s feature of different tanks＇ liquid level, this method can set high and low threshold automatically, consequently, setting a threshold artificially with the conventional Canny operator could be avoided. So this method can avoid unreasonable edge information, especially for the loss of liquid level＇s information because of improper artificial selection. In the process of extracting liquid level＇s information, it can analyze the number of transverse stripes, and obtain their idiographic position according to the number, making the edge operator more practical in the application of tanks＇ liquid level detection. The measurement results indicate that the error of liquid level is within 0.02m, and its expectation of proportional error is 0.062%.