Track abnormal condition detection method based on noise and image information fusion

The invention belongs to the technical field of audio processing, and discloses a track abnormal condition detection method based on noise and image information fusion, which uses a camera and a sound sensor to acquire image information and audio signals, thereby facilitating a system to judge track abnormal conditions more accurately. And track abnormal condition detection and positioning are combined, a track with abnormal conditions is positioned and uploaded to a track health management platform, related personnel need to be timely dispatched to carry out maintenance and recording subsequently, and finally, the related personnel need to be submitted to experts to review whether the maintenance is completed or not. 本发明属于音频处理技术领域，公开了一种基于噪声和图像信息融合的轨道异常情况检测方法，使用相机和声音传感器获取图像信息和音频信号，这样更有利于推动系统对于轨道异常情况更准确的判断。并将轨道异常情况检测与定位结合，对存在异常情况的轨道进行定位，并上传至轨道健康管理平台，后续需及时派遣相关人员进行维修与记录，并最终交与专家评审此次维修是否完成。