图像目标识别中的分层检索模型

Delamination Retrieval Model in Image Object Recognition

形状特征是武器装备目标识别的关键因素。针对归一化后轴夹角的特点和小波描述子多尺度描述轮廓曲线的特性,提出一种R树索引的分层检索技术模型,在模型库空间逐级过滤出与目标相似的模板,以保证检索的效率和准确性,并通过对军用车辆的检索实验验证了模型的有效性。

The shape features of military weapons & equipment are the most important key factors of image object recognition. Aiming at the characteristics of the contour-curve′s normalized rear-axes-angle and multi-scale description of the Wavelet Descriptor, this paper proposes a delaminating image retrieval model by using R-tree index, which hierarchically filter the similar templates with the object pattern via the multi-resolution description techniques of the contour to ensure the system′s efficiency and precisi...