【摘要】 针对煤粉输送管道泄露常发的问题，提出基于红外热像图及可见光图像共同检测的方法。该方法依据煤粉泄露烟雾的高温特性，利用红外热像仪监测现场温度，检测异常高温区域；同时，根据煤粉烟雾图像的特征，利用HSV颜色空间提取颜色特征，并采用灰度共生矩阵进行纹理特征提取，得到特征向量，再将特征向量输入BP神经网络进行识别。研究结果表明，泄露煤粉的高温能够维持较长距离，可以被红外热像仪检测；基于HSV颜色特征以及纹理特征能够有效识别煤粉烟雾，准确率达99.6%,为快速准确的煤粉输送管道泄露检测提供了保障。更多还原

【Abstract】 Targeting at the frequent leakage in coal fines conveying pipelines, a collaborative detection method based on infrared thermal images and visible images is proposed. Based on the high-temperature characteristics, an infrared thermal imager is used to monitor the on-site temperature and detect abnormal high-temperature areas. Based on the characteristics of coal fines smoke images, HSV color features are obtained, texture features are extracted by using the gray level co-occurrence matrix, and the feature vectors are input into a BP neural network for recognition. The research results demonstrate that the high temperature of leaked coal fines can be maintained for a long distance and can be detected by infrared thermography. Based on HSV color and texture features, coal fines smoke can be effectively identified with an accuracy rate of 99.6%,providing a guarantee for rapid and accurate leakage detection of coal fines conveying pipelines.更多还原