Few-shot　class-incremental　learning　(FSCIL)　means　identifying　new　classes　in　a　few　samples　while　not　forgetting　the　old　ones.　The　challenge　of　this　task　is　that　new　class　has　only　few　supervised　information　during　the　learning　process.　Aiming　to　boost　the　performance　of　FSCIL,　we　propose　a　novel　method　in　this　paper.　To　be　clear,　our　method　has　two　contributions　as　follows.　First,　we　elegantly　employ　the　principal　component　analysis　(PCA)　and　adopt　a　model　with　a　strong　prior　for　feature　extracting,　specifically,　we　decouple　the　feature　extractor　from　classifier　in　the　incremental　learning　process.　Second,　we　innovatively　introduce　the　data　augmentation　during　the　learning　process　of　FSCIL　to　enhance　the　sample　diversity　and　get　a　more　accurate　class　prototype　based　on　enriched　samples.　Excellent　experimental　results　on　CIFAR100,　miniImageNet,　and　CUB200　datasets　verify　the　superiority　of　our　method,　compared　to　several　existing　methods.