论文部分内容阅读
针对实际红外空中图像中目标的大小、方向等发生变化等情况 ,基于奇异值分解变换的特征向量稳定性不好。为解决这一技术难题 ,提出了基于尺度奇异值变换的红外图像目标特征提取方法 ,该方法是一种基于尺度变换的矩阵表示方法。同类目标图像样本经过上述变换后所得到的矩阵都具有相似的能量 ,用文中提出的矩阵表示方法对红外图像进行描述后 ,对其对应的奇异值进行特征提取 ,克服了对原始图像的矩阵进行奇异值特征提取时存在的缺点 ,实验表明该方法提取目标矩阵的奇异值特征向量具有较好的稳定性 ,对图像的平移、旋转和比例变化具有良好的不变性 ,是一种实用的特征提取方法。
In view of the fact that the size, direction and the like of the target in the real infrared aerial image change, the stability of the eigenvector based on the singular value decomposition transform is not good. In order to solve this technical problem, an infrared image feature extraction method based on scaling singular value transform is proposed, which is a matrix representation method based on scale transformation. Similar matrices of the target image samples after the above transformation have similar energies. After the infrared images are described by using the matrix representation method presented in the text, the corresponding singular values are extracted by feature extraction, which overcomes the problem that the matrices of the original images The experimental results show that the singular value eigenvector extracted from the target matrix has good stability and good invariance to the image translation, rotation and scale change, and is a practical feature extraction method.