基于仿射不变离散哈希的遥感图像快速目标检测新方法

A novel fast object detection method in remote sensing image based on affine-invariant discrete hashing

提出一种基于仿射不变离散哈希的遥感图像快速目标检测新方法.首先使用一种"选择性搜索"的方法生成目标候选框;其次,提出一种基于仿射不变离散哈希(Affine-Invariant Discrete Hashing,AIDH)的目标检测方法,该方法采用具有低存储、高效率优势的监督离散哈希框架,结合仿射不变优化因子,构造仿射不变离散哈希,通过将具有相同语义信息的仿射变换样本约束到相似的二值码空间,实现检测精度的提高;最后采用判别分类器结合非极大值抑制的方法,进一步过滤掉误检目标框,完成目标的精确定位.实验证明,在NWPU VHR-10数据集下,该方法相比于经典目标检测方法和新的哈希方法,在具备高效性的同时,在精度上也得到了保证.

A novel fast object detection method in remote sensing images based on Affine-Invariant Discrete Hashing(AIDH)is proposed.Firstly,the selective search method is introduced for region proposal generation.Secondly,affine-invariant discrete hashing is suggested for object detection.This method uses supervised discrete hashing with the advantage of low storage and high efficiency,jointed with affine-invariant factor,to construct affine-invariant discrete hashing.By constraining the affine transform samples with the same semantic information to the similar binary code space,the method achieves the enhancement on classification precision.Finally,we use discriminating classifier and non-maximum suppression for further filtering false detection of object area and accomplishing accurate object localization.Experiments show that under the dataset of NWPU VHR-10 the proposed method is more efficient than classic detection method and new hash method,and it is also guaranteed in accuracy.