Sea target size detection method based on vision and laser sensor data fusion

Abstract

The invention relates to a sea target size detection method based on vision and laser sensor data fusion Firstly joint calibration is performed on the laser radar and the camera; then recognition training is performed on the sea target by using the visual SSD detection algorithm so that the camera can accurately select the obstacles at sea; the laser point cloud data detected by the laser are projected onto a two-dimensional raster map and the point cloud in the raster map is clustered so as to obtain the total number n of the obstacles on the sea surface detected by the laser; as for the obstacles with a distance less than 80 meters, the obstacle information detected by the laser radar counts; and as for the obstacles with a distance greater than 80 meters, the information of laser and vision fusion detection is used as the final detection result. According to the method, the approximate size information of the obstacles within the range of 200 meters and the distance from the obstacle to the unmanned ship can be obtained so as to greatly make up the defect of single sensor and improve the stability and the accuracy of target detection.

Description

technical field [0001] The invention relates to a method for detecting the size of a sea target based on vision and laser sensor data fusion, and belongs to the technical field of unmanned boat applications. Background technique [0002] Countries all over the world are conducting more and more research on the application of surface unmanned vehicles, and the requirements for the intelligence level of surface unmanned vehicles are also constantly increasing. Unmanned surface vehicles have the ability to collect useful information from the surrounding environment, construct a symbolic environment model, and use the model to plan and execute upper-level tasks, among which maritime environment perception is one of the core issues. [0003] Maritime environment perception refers to the use of existing sensors, such as cameras, laser radars, and maritime radars, to obtain the water surface conditions within hundreds of meters or even kilometers around the unmanned boat, and to ob...

AI Technical Summary

Problems solved by technology

[0004] When the existing environmental perception sensors are used alone, there are some shortcomings, and it is difficult to meet the actual requirements of unmanned boat navigation

For monocular vision, it can identify various obstacles on the sea through the camera, but it cannot accurately determine the distance between a certain obstacle and the unmanned boat and the size of the obstacle itself; for radar, It can get the distance and speed information of the target at sea, but it cannot get the specific size of the obstacle, nor can it distinguish the type from the shape of the obstacle; while for the lidar, although it can get the approximate shape and shape of the surrounding obstacles It can also know the distance between the target and the ship, but due to the limitation of the detection distance of the laser sensor itself, when the detection distance is greater than 150 meters, its point cloud becomes very sparse, and it is impossible to accurately obtain what we want. Obstacle information

Images