In the task of detecting targets under airport pavement， B-scan maps generated by Ground Penetrating Radar （GPR） have complex backgrounds and lots of noise， especially a single B-scan map cannot reflect the complete information of an underground target. To solve these problems， a Three-Dimensional Channel and Spatial Attention UNet （3D-CSA-UNet） model was established to automatically detect the underground targets. Firstly， a Three-Dimensional Channel and Spatial parallel attention Block （3D-CS-Block） was designed to make the model focus on the underground target information in radar C-scan and suppress the interference of backgrounds and noise. Secondly， in order to enhance the capability of 3D-CS-Block in feature extraction， a multi-scale 3D segmentation model was designed to extract feature maps of different sizes from the radar C-scan. Finally， the cross-entropy loss function was employed to calculate the loss value of feature map under each scale to improve the detection accuracy of the model. On a real dataset of targets under airport pavement， compared with 3D-Fully Convolutional Network （3D-FCN）， 3D-UNet and other algorithms， 3D-CSA-UNet has the average F1 score in terms of the pixel level segmentation for void， rebar and parallel rebar targets increased by at last 12.33， 9.05 and 11.05 percentage points. Experimental results show that 3D-CSA-UNet can meet the real engineering requirements well.

To solve the problems of high-dimensionality, giant-computation and redundancy of Gabor features in current Facial Expression Recognition (FER), a new FER algorithm based on Gabor Parameter Matrix (GPM) and improved Adaboost was proposed. Firstly, the GPM was defined by combining pixel information of image and parameters of Gabor wavelet kernel; Secondly, the idea of Genetic Algorithm (GA) was introduced into Adaboost to improve its searching performance, then the improved Adaboost was used to select optimal features corresponding to the elements in GPM to build strong classifiers, thereby the dimensionalities, redundancy and calculation amount of Gabor features could be reduced by feature selection; Finally, on the basis of building several strong classifiers, a multi-expressions classification algorithm was developed to implement FER. The experimental results on Matlab indicate that average expression recognition rate of the proposed algorithm is 89.67%, and the selection efficiency of optimal features is improved significantly.

The security of E-commerce determines its development; however, traditional means of security certification cannot meet the demand on reliability. To ensure the security in E-commerce communication, a new method of security certification was proposed based on rolled fingerprint reconstruction and digital signature. Firstly, the image sequence of rolling fingerprints was reconstructed into a high-quality complete fingerprint image, and then the features were extracted as the key of digital signature. Thirdly, the message embedded with digital signature was transferred. At the same time, the correctness and completeness of the message were checked, and the validity of the sender was also identified. The experimental results show that the proposed algorithm can be used for rolling fingerprint reconstruction in any direction to get a high-quality fingerprint image. And it can run in real-time because of low complexity. The rolling fingerprint, as the security certification medium, ensures the security of E-commerce.