Abstract Realistic images generated by GANs (generative adversarial networks) have enriched people’s lives but also pose serious threats to personal privacy and society. Therefore, it has become essential to develop methods for accurately detecting GAN-generated images. Existing methods have utilized artifacts to detect GAN-generated images, but the artifacts in different generated images can vary significantly. As a result, these algorithms often struggle to detect generated images whose artifacts differ substantially from those in the training set. This paper proposes the PixMSE algorithm based on statistical features, which designs the MSENet1 process to obtain local roughness feature maps of the image and the MSENet2 process to obtain local roughness feature maps of the filtered image. Finally, PixMSE uses features extracted by ResNet (Residual Network) to detect the generated images. Experimental results show that PixMSE achieves an average detection accuracy of 84.3% on eight subsets of the Wang dataset, which is a publicly available dataset, demonstrating strong cross-model generalization performance. The source code and datasets are available at https://github.com/DarlingDiving/PixMSE.

In infrared imaging, spatially proximate small targets undergo severe aliasing due to the point spread function (PSF), forming blurred, indistinguishable patterns. This paper proposes STSA-Net, a structure-enhanced deep unfolding network for separating and reconstructing closely-spaced infrared small targets (CSIST). Built upon the DISTA-Net framework, the method introduces Small Target Sensitive Convolution (STSC) to enhance local peak responses of point-like sources and designs an Adaptive Sparse Threshold (AST) module that combines global statistics with spatial modulation to generate pixel-wise thresholds, thereby achieving robust target-background separation. On the CSIST-100K dataset, STSA-Net outperforms DISTA-Net by 0.32 and 1.03 percentage points in mAP and AP-20, respectively, while maintaining comparable reconstruction quality. Code: https://github.com/UNCERTAIN-FATE/STSA-Net.

Happiness is a hot topic in academic circles. The study of happiness involves many disciplines, such as philosophy, psychology, sociology, and economics. However, there are few studies on the quantitative analysis of the factors affecting happiness. In this article, we used the well-known World Values Survey Wave 6 (WV6) dataset to quantitatively analyze the happiness of 57 countries with Big Data techniques. First, we obtained the seven most important factors by constructing happiness decision trees for each country. Calculating the frequencies of these factors, we obtained the 17 most important indicators for the prediction of happiness in the world. Then, we selected five representative countries, namely, Sweden, Japan, India, China, and the USA, and analyzed the indicators with the random forest method. We identified different patterns of factors that influence happiness in different countries. This study is a successful attempt to apply data mining technology in the social sciences, and the results are of practical significance.