摘要
目标在成像过程中发生的几何变形在更多情形时用仿射变换来刻画,光照变化也是影响目标识别的重要因素.在一种典型的简化的光照变化约束下,将光照的灰度线性变化融入目标的几何变形,基于子空间流形理论,提出了仿射光照不变形状空间概念,分析了仿射光照不变形状空间的非线性几何结构,给出了仿射光照不变形状识别算法,模拟图像和真实图像序列实验验证了算法的有效性,该几何框架为研究几何和光照变化下的目标识别提供了新的途径.
The geometric warps of the object in the imaging process should be represented by the affine transformation at most situations. The illumination variation is one of the most important factors to affect the object recognition. Under the condition of a representative and simplified illumination variation, we syncretize the gray linear variation and the object geometric warps, propose the concept of affine illumination invariant shape space, analyze its nonlinear geometry structure and present the affine illumination invariant shape recognition algorithm using the subspace theory. Experiment results on the simulated and real images verify the efficiency of the proposed framework and algorithm. This geometric framework provides a novel approach to study the automatic target recognition under the illumination variation.
出处
《广州航海学院学报》
2016年第3期43-46,共4页
Journal of Guangzhou Maritime University
基金
河南省高等学校青年骨干教师资助计划(2012GGJS-297)
河南省科技攻关项目(142102210366)
关键词
形状识别
子空间流形
仿射变换
光照不变
shape recognition
subspace manifold
affine transformation
illumination invariant
