基于双目视觉的核相关滤波目标检测跟踪算法  

Kernelized Correlation Filters Object Detection and Tracking Algorithm Based on Binocular Vision

作　　者：陶利民 张婧 韩伟 唐涛 蒋铋铖 TAO Limin;ZHANG Jing;HAN Wei;TANG Tao;JIANG Bicheng(College of Mechanical,Naval Architecture and Ocean Engineering,BeiBu Gulf University,Qinzhou 535011,China)

机构地区：[1]北部湾大学机械与船舶海洋工程学院,钦州535011

出　　处：《船舶》2025年第2期95-102,共8页Ship & Boat

摘　　要：针对海洋工程领域技术不断创新,以及对提高适应复杂海况能力需求持续增长的现状,该文以波浪补偿装置检测系统为背景,运用基于双目视觉的核相关滤波(kernelized correlation filters,KCF)算法作为目标检测方法。该方法基于双目视觉检测负载运动,获取图像上的特征点;并通过坐标系变换求解负载在惯性坐标系中的位置,实时检测被补给物的空间位置和姿态,从而通过对负载六自由度运动的控制,使其与被补给舰无相对运动,实现波浪补偿;基于机器人操作系统(robot operating system,ROS)在Ubuntu系统下进行编程,通过波浪补偿样机进行对比及仿真实验,验证所采用的检测跟踪方法可行性。结果表明,该方法可很好地适用于波浪补偿系统。In view of the continuous technological innovation in marine engineering and the growing demand for enhanced adaptability to complex sea conditions,the binocular vision-based KCF(Kernelized Correlation Filters)algorithm is used as the target detection method based on the wave compensation device detection system.Based on the binocular vision,this method detects the load motion,capture the image feature points,calculates the position of the load in the inertial reference frame through coordinate system transformation,and monitors the spatial position and attitude of the replenished object in real time.The relative movement between the load and the replenishment ship can be eliminated by controlling the six-degree-of-freedom motion of the load,thereby achieving wave compensation.Program is developed under the Ubuntu system based on the robot operating system(ROS).Comparison and simulation are then conducted with a wave compensation prototype to validate the feasibility of the proposed detection and tracking method.The results indicate that this method is suitable for wave compensation systems.

关 键 词：双目视觉 KCF算法 目标检测 位姿解算 

分 类 号：TS736[轻工技术与工程—制浆造纸工程]