自适应加权特征字典与联合稀疏相结合的遥感目标检测

Remote Sensing Targets Detection Based on Adaptive Weighting Feature Dictionaries and Joint Sparse

随着分辨率的提高,遥感图像空间包含的有用信息越来越丰富,这使得遥感数据的处理变得更加复杂,容易发生维数灾难并影响识别效果。针对这一情况,提出一种自适应加权特征字典与联合稀疏相结合的遥感图像目标检测方法(GJ-SRC)。首先将训练图像和待测图像进行Gabor变换以提取特征图像。然后计算各个特征值在进行稀疏表示时的贡献权重,通过自适应方法构造特征字典,使字典具有更强的判别能力。最后,提取每一类图像的公共特征和单个图像的私有特征构成联合字典,并利用测试图像稀疏表示进行目标检测识别。为了避免Gabor变换产生的维数灾难,在处理过程中采用PCA方法对特征字典进行降维,以降低计算成本。实验表明,与现有的SRC方法和遥感目标检测方法等相比,所提方法具有较好的检测效果。

With the improvement of resolution,more and more useful information is contained in remote sensing images,which makes the processing of remote sensing data become more complex,and it is easy to cause the curse of dimensionality and the poor recognition effect.In view of this situation,a remote sensing targets detection approach(GJ-SRC)based on adaptive weighting feature dictionaries and joint sparse was proposed.Firstly,the Gabor transform is used to extract the features from the training images and the testing images.Then,the contribution weights of each eigenvalue in sparse representation are calculated,and the feature dictionary is constructed by the adaptive method,which makes the dictionary more discriminative.Finally,the common features of each category and the private features of a single image are extracted to form a joint dictionary,and the sparse representation of the test image is used for target recognition.In order to avoid the curse of dimensionality caused by the Gabor transform,the PCA method is used to reduce the dimension of the feature dictionary in order to reduce the computational cost.Experiments show that this method has better detection effect compared with the existing SRC method and remote sensing target detection method.