The invention discloses an image recognition method applied to unmanned aerial vehicle routing inspection of an expressway emergency lane. The method comprises the following steps of S1, making a video shot by a shooting device into an independent video segment according to a set duration; S2, calling a video segment; S3, extracting at least one frame of image in the video segment; S4, performingsemantic segmentation on the extracted image, and identifying a lane area and a background area from the image; S5, performing lane curve fitting on the lane area, and simulating a travel lane line and an emergency lane line; S6, judging whether a vehicle exists in the emergency lane or not through pattern recognition, if so, executing S7, and if not, executing S8; S7, communicating with a groundstation, and transmitting the video segment to the ground station; and S8, deleting the video segment. On the one hand, only part of the frame pictures are extracted, the work intensity is low, few resources are occupied; on the other hand, data transmission does not need to be conducted on the whole video, the transmission cost is reduced, and energy losses in the data transmission process are reduced.
本发明公开了一种应用于高速公路应急车道无人机巡检的图像识别方法,包括以下步骤:S1、将拍摄装置拍摄的视频按设定时长制成独立的视频段;S2、调用视屏段;S3、提取视频段中的至少一帧图像;S4、对提取到的图像进行语义分割,从该图像上识别出车道区域和背景区域;S5、对车道区域进行车道曲线拟合,模拟出行车道线和应急车道线;S6、通过图形识别,判断应急车道内是否有车辆,如有则执行步骤S7,如无则执行步骤S8;S7、与地面站台通信,将上述视频段传输至地面站台;S8、删除上述视频段。本发明一方面只提取部分帧图片,工作强度低、占用资源少,另一方面,无需对整个视频进行数据传输,也降低了传输成本,降低了数据传输过程中的能源损耗。
Image recognition method applied to unmanned aerial vehicle routing inspection of expressway emergency lane
一种应用于高速公路应急车道无人机巡检的图像识别方法
2021-01-22
Patent
Electronic Resource
Chinese
Multifunctional unmanned aerial vehicle expressway inspection system
European Patent Office | 2023
|Multifunctional routing inspection unmanned aerial vehicle
European Patent Office | 2021
|