Aerial Image Classification Using Color Coherence Vectors and Rotation & Uniform Invariant LBP Descriptors

Taking texture features as basis and combining color coherence vector, a fusion classification algorithm employed rotation & uniform invariant LBP and color coherence vector was proposed for optical aerial images. Firstly, rotation & uniform invariant texture feature and color coherence vector feature were extracted. Then, feature weighted fusion was performed according to the minimum misjudgment rate of negative samples. At last, Support Vector Machine (SVM) based on RBF was applied to classify image. The experimental results on NWPU-RESISC45 data sets reach 96.66%, which indicate that the algorithm is an effective aerial image classification method.