Laser & Optoelectronics Progress, Volume. 52, Issue 7, 71501(2015)

An Affine Invariant Method of Forward Looking Infra-Red Target Recognition

Zhao Aigang*, Wang Hongli, Yang Xiaogang, Lu Jinghui, Wang Jianyong, and Cui Xiangxiang
Author Affiliations
  • [in Chinese]
  • show less

    Forward looking infrared target recognition is key technology in precision-guided weapon. In practical application, it is likely to cause wrong matching because of discrepancy of scale and angle between template and real images. In order to solve problem mentioned above, A forward looking infrared target recognition algorithm based on edge′s affine invariant coordinate. Establishing coordinate system based on regional centroids of curvic edge. Coordinates of edge points is affine invariant in the established coordinate system. Taking the set of edge points′ coordinates as edge descriptor and similarity of edge is measured by average hausdorff distance. Wrong matchings are excluded with the help of relative locations among edges and forwarding looking infrared target robust recognition is implemented. Compared with two recognition algorithms based on gray template and edge template, recognition rate and robustness of proposed algorithm are all improved through three videos.

    Tools

    Get Citation

    Copy Citation Text

    Zhao Aigang, Wang Hongli, Yang Xiaogang, Lu Jinghui, Wang Jianyong, Cui Xiangxiang. An Affine Invariant Method of Forward Looking Infra-Red Target Recognition[J]. Laser & Optoelectronics Progress, 2015, 52(7): 71501

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Machine Vision

    Received: Nov. 4, 2014

    Accepted: --

    Published Online: Jul. 6, 2015

    The Author Email: Aigang Zhao (zhaoaigang1986120@163.com)

    DOI:10.3788/lop52.071501

    Topics