Abstract: A hyperspectral image target detection method based on sparse representation with neighborhood weighted is proposed. In the construction of a sparse model, the similarity of pixels is represented by the inner product of the unit pixel and the constructed image is dealt with neighborhood weighted constraints, which can provide smooth space. Furthermore, orthogonal matching pursuit algorithm based on weighted least squares is proposed to solve the problem. It can ensure the effectiveness of the parameter. The experimental results show that detection algorithm in this paper is effective and feasible.

Key words: hyperspectral image, target detection, sparse representation, neighborhood weighted

摘要： 提出了一种基于邻域加权稀疏表示的高光谱图像目标探测方法。在构造稀疏模型时，以单位化像元的内积表示像元的相似性，据此对重构图像中测试像元空间邻域的像元进行加权约束，保证了空间的平滑性；并提出基于加权最小二乘的正交匹配追踪算法求解该稀疏模型，它使得每次迭代中参数估计有效。实验结果表明，该探测算法是有效可行的。

关键词: 高光谱图像, 目标探测, 稀疏表示, 邻域加权