A multi-sensor decision-level fusion method for intelligent ship surface target perception and recognition

Abstract

The invention relates to the field of unmanned intelligent ships, in particular to a multi-sensor decision-level fusion intelligent ship surface target perception and identification method, comprising an optical camera and a plurality of sensors. Any two or more of , including the following steps, firstly calibrate the internal and external parameters of the sensor; then, according to the target results detected by the navigation radar ARPA target, the AIS target and the lidar, a matching fusion method of multi-sensor target decision-level fusion is designed. The position of the water surface target in the world coordinate system is obtained; finally, the azimuth angle coincidence degree of the image recognition result of the forward-looking camera is used to match the recognition result of the fusion target and the forward dangerous target category. Through the complementary advantages of each sensor, highly robust environment perception and obstacle detection and recognition can be obtained, which provides guarantee for the autonomous navigation of smart ships, collision avoidance and target tracking.

Description

technical field [0001] The invention relates to the field of unmanned intelligent ships, in particular to a multi-sensor decision-level fusion intelligent ship surface target perception and identification method. Background technique [0002] As a scalable surface task platform, the surface unmanned ship has a complex working environment and a high degree of unknown area. All-weather, long-term, highly adaptable, and highly robust automatic target detection, recognition, and tracking are the key technologies to ensure their safe navigation and mission execution. When the unmanned ship platform performs environmental perception and target recognition, the main means used are radar, AIS and electro-optical imaging equipment, but these methods have some limitations. [0003] The first is the adaptability of the sea surface environment. First, the typical feature of small tonnage unmanned ships compared with other unmanned intelligent systems is that the platform is highly dyna...

AI Technical Summary

Problems solved by technology

[0003] The first is the adaptability of the sea surface environment. First, compared with other unmanned intelligent systems, the typical characteristic of small-tonnage unmanned ships is that the platform is highly dynamic in high sea conditions, and the target shakes out of the detection range of the sensor, causing detection and tracking failures, especially Elevation changes affect the target distance calculation accuracy of the optical system; second, the sea environment is complex and changeable, and the probability of rain and fog is high, which affects the detection effect of the sensor and makes it difficult to analyze and judge the image of the optical equipment

[0004] Secondly, there are limitations in the sensor itself. The navigation radar has a long detection range but there is a blind spot in the short range, the detection frequency is low, and it is difficult to capture and track small and fast targets. The photoelectric system has high resolution and can distinguish the characteristics of the target, but the target ranging error is relatively large Large, and there are limitations in the ability to adapt to the environment; LiDAR can obtain accurate distance information and three-dimensional contour information of the target, but it does not have long-distance detection capabilities, vertical resolution, and it is difficult to unify targets between time frames in high sea conditions

[0005] At present, there are few researches on multi-sensor collaboration and information fusion, and there is still a lack of mature methods

Images