ConvNet [19] is essentially a kind of input-to-output mapping. The purpose of the training is to learn the mapping
ability between input and output. The model of the ConvNet is mainly composed of several numbers of convolutional layers and pooling layers. Finally use high-level features, which generated by fully connected layers, as the basis of identifying the input image. The convolutional layer make use of convolutional kernels to process convolutional evaluation with the input image. These image features then convey to the next layer as mapping feature. Thus the convolutional layer is just an image convolution of the previous layer.