Multi-sensor fusion and personnel positioning method based on image processing

Abstract

The invention discloses a multi-sensor fusion and personnel positioning method based on image processing. The multi-sensor fusion and personnel positioning method comprises the following steps: S1, acquiring a three-channel RGB color image and a one-channel multi-line laser ranging image; S2, projecting camera optical coordinates of the RGB color image into laser point cloud coordinates, and projecting the laser point cloud coordinates into 360-degree annular panoramic coordinates; S3, performing target detection frame selection on a pre-trained target by using a deep learning image recognition technology to obtain a target detection bounding box distribution image and an object category distribution image; S4, searching a laser point cloud coordinate of a central point of the target detection frame as a laser ranging value of the target; S5, calculating a monocular distance measurement value of the camera through the target detection frame; and S6, carrying out Kalman filtering on thelaser ranging value and the monocular distance measurement value, and obtaining the final target distance value. The multi-sensor fusion and personnel positioning method is simple, high in real-timeperformance and capable of rapidly and accurately achieving transformer substation personnel positioning.

Description

technical field [0001] The invention relates to the field of substation personnel positioning, in particular to a multi-sensor fusion and personnel positioning method based on image processing. Background technique [0002] The substation is an important part of the stable operation of the power system. Various important substation equipment are distributed in the substation. Although the existing substation is generally a closed area, outsiders cannot enter the substation, but for the staff in the substation. It is difficult to effectively control the behavior of the staff, especially in some areas in the substation, it is difficult to monitor accurately with the existing camera monitoring, and there are blind spots in the camera monitoring. More importantly, the substation There are many types of equipment and the structure is not fixed. Compared with the line, the maintenance operation environment is more complicated, and the maintenance operation process is more cumberso...

AI Technical Summary

Problems solved by technology

[0002] The substation is an important part of the stable operation of the power system. Various important substation equipment are distributed in the substation. Although the existing substation is generally a closed area, outsiders cannot enter the substation, but for the staff in the substation. It is difficult to effectively control the behavior of the staff, especially in some areas in the substation, it is difficult to monitor accurately with the existing camera monitoring, and there are blind spots in the camera monitoring. More importantly, the substation There are many types of equipment and the structure is not fixed. Compared with the line, the maintenance operation environment is more complicated, and the maintenance operation process is more cumbersome. Therefore, the safety distance between personnel, equipment and equipment should be regulated during maintenance operations

[0003] At present, the positioning of substation operation and maintenance personnel mostly uses technologies such as camera shooting and laser radar positioning. Manual detection is used for personnel identification on the captured images, which is time-consuming and laborious, or image processing is used for personnel positioning, but the existing problems are cameras and Both lidar lenses are spherical lenses, but the coordinate scale lines of the two are completely different, resulting in different directions of barrel distortion

However, this processing method is complex in calculation and poor in real-time performance. At the same time, the workload of personnel positioning tracking and target distance detection on a single layer on the entire image is huge, the positioning effect is poor, and the accuracy of distance calculation is low.

Images