The bidirectional 2DPCA (two-dimensional principal component analysis) method for SAR images recognition, can compress the columns and rows of images matrix and reduce the number of feature dimensions. However, it fails to use high order statistics information of image data, neglects the nonlinearity correlation between pixels. Therefore, this paper presents the method combined bidirectional 2DPCA with KPCA (Kernel Principal Component Analysis). This method not only compresses the dimensions of images data, but develops the superiority of KPCA in describing correlation between many pixels. Experimental results show that: this method can decrease calculated amount and raise recognition rate of SAR target effectively.