基于红外视觉/激光雷达融合的目标识别与定位方法

近年来,基于可见光图像的目标识别在无人车感知领域得到了广泛应用.然而,可见光图像目标识别无法应用于弱光和黑暗环境.针对于此,提出了一种基于红外视觉/激光雷达融合的目标识别与定位算法.首先,通过基于颜色迁移的数据增强训练方法,提高了红外目标识别算法的泛化性能.继而,提出了一种基于激光雷达修正的单目深度估计方法,通过视觉图像与激光雷达点云的数据融合,实现了基于稠密深度图对目标位置的估计,提高了对小目标的定位能力.试验结果表明,与传统Yolov3目标识别算法相比,该算法平均识别准确率可提升5.8％;此外,相对定位算法将包含小目标在内的物体相对定位精度提高了13.4％.

Object Detection and Positioning Method Based on Infrared Vision/Lidar Fusion

In recent years, object detection based on visible light images has been widely used in the field of unmanned vehicle perception. However, visible light image target recognition cannot be applied to low light and dark environments. In response to this, the paper proposes a object detection and positioning algorithm based on infrared vision/lidar fusion. First, the generalization performance of the infrared object detection algorithm is improved through the data enhancement training method based on color migration. Then, a monocular depth estimation method based on lidar correction is proposed. Through the fusion of visual image and lidar point cloud data, the position estimation of the target based on the dense depth map is realized, and the localization ability of small objects is improved. The test results show that compared with the traditional Yolov3 object detection algorithm, the average detection accuracy of the proposed algorithm can be increased by 5.8%; In addition, the relative positioning algorithm improves the relative positioning accuracy of objects including small targets by 13.4%.