一种改进的转炉出钢下渣检测方法及其应用

 针对传统的完全基于灰度阀值分割算法的转炉出钢下渣检测方法在遇到出钢过程中的意外情况时，检测准确率下降的缺点，提出了一种改进的红外转炉出钢下渣检测方法。该方法首先利用分水岭分割的方法将转炉出钢图像分割成几个区域，再利用形态学方法对分割出的区域进行矩形拟合以获取准确的钢流图像区域，最后通过前后共4帧图像信息的加权来消除出钢过程中的异常情况对于检测准确性的影响。现场试验结果表明，所提出的改进方法对于提高红外转炉出钢下渣检测系统报警准确率，减小下渣到钢包的渣厚度以及提高系统稳定性方面效果显著。

An Improved Converter Tapping Slag Detection Method and Its Application

The traditional converter tapping slag detection method was based entirely on threshold segmentation, however, in the face of accidents occurred during converter tapping, the detection accuracy dropped. In order to solve such shortcoming, an improved converter tapping slag detection method was put forward. First, converter tapping image was divided into several areas by using watershed segmentation method. Then, rectangular fitting method was used to obtain steel image area. Finally, the impact of anomalies occurred during converter tapping was eliminated through weighting 4 images′ information. Industrial test results show that the method is effective to improve accuracy and stability of infrared slag detection system and control thickness of slag in ladle.