Multi-label image classification is an essential task in image processing. How to improve the correlation between labels by learning multi-scale features from images is a very challenging problem. We propose a Double Attention Network (DAN) to improve the correlation between the image feature regions and the labels, as well as between labels and labels. Firstly, the dynamic learning strategy is used to extract the multi-scale features of the image to solve the problem of inconsistent scale of objects in the image. Secondly, in order to improve the correlation between the image feature regions and the labels, we use the spatial attention module to focus on the important regions of the image to learn their salient features, while we use the channel attention module to model the correlation between the channels to improve the correlation between the labels. Finally, the output features of two attention modules are fused as the multilabel image classification model. Experiments on MS-COCO 2014 dataset, Pascal VOC 2007 dataset and NUS-WIDE dataset demonstrate that our model is significantly better than the state-of-the-art models. Besides, visualization analyses show that our model has a strong ability for image salient feature learning and label correlation capturing.
Visual odometry (VO) estimation from blurred image is a challenging problem in practical robot applications, and the blurred images will severely reduce the estimation accuracy of the VO. In this paper, we address the problem of visual odometry estimation from blurred images, and present an adaptive visual odometry estimation framework robust to blurred images. Our approach employs an objective measure of images, named small image gradient distribution (SIGD), to evaluate the blurring degree of the image, then an adaptive blurred image classification algorithm is proposed to recognize the blurred images, finally we propose an anti-blurred key-frame selection algorithm to enable the VO robust to blurred images. We also carried out varied comparable experiments to evaluate the performance of the VO algorithms with our anti-blur framework under varied blurred images, and the experimental results show that our approach can achieve superior performance comparing to the state-of-the-art methods under the condition with blurred images while not increasing too much computation cost to the original VO algorithms.
24 July 2025 in Journal on Computing and Cultural Heritage
The rich and enduring traditional culture cultivated by the Chinese nation over millennia serves as the core and essence of the Chinese heritage. In this intricate tapestry of culture, traditional patterns hold a significant and revered position. These patterns not only reflect the aesthetic values and artistic achievements of ancient Chinese civilization but also carry deep cultural and historical significance. However, the segmentation and recognition of traditional patterns pose significant challenges due to the limited availability of annotated data, the complexity of pattern variations, and the interference of material textures. To address these challenges, this article proposes a traditional pattern segmentation algorithm based on the memory learning model. The memory learning model, as the guiding principle of the algorithm, leverages prior knowledge from related domains to enable the algorithm to generalize effectively with limited annotated data. The algorithm consists of two key components: a saliency prior module and a multi-scale feature matching module. The saliency prior module uses phase spectrum information to generate saliency maps, guiding the model to focus on high-frequency information such as edges and contours. The multi-scale feature matching module captures features at different scales, improving the robustness and accuracy of the segmentation. We construct a traditional pattern dataset by introducing a phase spectrum–amplitude spectrum fusion algorithm, which enhances the model’s ability to focus on phase consistency information. Experimental results on a traditional pattern dataset show that our proposed algorithm outperforms state-of-the-art methods, demonstrating its superior performance and robustness in handling complex and diverse pattern segmentation tasks.
9 December 2025 in IEEE Transactions on Computational Social Systems
Normal estimation is a critical task in point cloud analysis, especially in cultural heritage preservation and digitization. However, due to errors from acquisition devices and environmental noise, the normal information in point cloud data often suffers from inaccuracies. Therefore, accurate normal estimation and optimization of point cloud data are essential steps for improving the effectiveness of subsequent processing. To address these issues, we propose a novel point cloud normal re-estimation algorithm based on adaptive consistency propagation (NRACP) for cultural artworks. This method adaptively adjusts the selection of neighboring points to achieve consistent normal estimation across surfaces of varying complexity. Specifically, the NRACP algorithm dynamically determines the number of neighboring points based on the local geometric features of the point cloud. Then, it uses a consistent propagation mechanism to optimize the normal directions iteratively, ensuring the stability and accuracy of the normal estimation. The NRACP algorithm demonstrates a notable enhancement in normal estimation accuracy, achieving a reduction of 2.37% in RMSE specifically under the 0.6% noise condition compared to previous methods, and showing particular strength in regions characterized by complex surfaces and elevated noise levels.
Near-infrared (NIR) hyperspectral imaging has become a critical tool in modern analytical science. However, conventional NIR hyperspectral imaging systems face challenges such as high cost, bulky instrumentation, and inefficient data collection. In this work, we demonstrate a broadband NIR compressive spectral imaging system that is capable of capturing hyperspectral data that cover a broad spectral bandwidth ranging from 700 to 1600 nm. By segmenting wavelengths and designing specialized optical components, our design overcomes hardware spectral limitations to capture broadband data, while the reflective optical structure makes the system compact. This approach provides a novel technical solution for NIR hyperspectral imaging.