Junyao Xing

Dongba scripture sentence segmentation is an important and basic work in the digitization and machine translation of Dongba scripture. Dongba scripture sentence segmentation line detection (DS-SSLD) as a core technology of Dongba scripture sentence segmentation is a challenging task due to its own distinctiveness, such as high inherent noise interference and nonstandard sentence segmentation lines. Recently, projection-based methods have been adopted. However, these methods are difficult when dealing with the following two problems. The first is the noisy problem, where a large number of noise in the Dongba scripture image interference detection results. The second is the Dongba scripture inherent characteristics, where many vertical lines in Dongba hieroglyphs are easily confused with the vertical sentence segmentation lines. Therefore, this paper aims to propose a module based on the convolutional neural network (CNN) to improve the accuracy of DS-SSLD. To achieve this, we first construct a tagged dataset for training and testing DS-SSLD, including 2504 real images collected from Dongba scripture books and sentence segmentation targets. Then, we propose a multi-scale hybrid attention network (Multi-HAN) based on YOLOv5s, where a multiple hybrid attention unit (MHAU) is used to enhance the distinction between important features and redundant features, and the multi-scale cross-stage partial unit (Multi-CSPU) is used to realize multi-scale and richer feature representation. The experiment is carried out on the Dongba scripture sentence segmentation dataset we built. The experimental results show that the proposed method exhibits excellent detection performance and outperforms several state-of-the-art methods.

With the rapid development of outdoor vision system, removing rain streaks from a single image has attracted considerable attention as rain streaks can affect the quality of the image taken in rainy days, and interrupt the key information, which will greatly reduce the use value of the image, thus affecting the performance of traffic, safety monitoring and other facilities. Although the deep learning methods have achieved satisfying performance in single image de-raining, there are still two problems: First, the rain streaks contained in one dataset we can use are limited, and in the case of real rainy days, the rain streak density is diverse, it is impossible to accurately classify them. Therefore, the existing rain removal models cannot remove rain streaks properly for images with different rain streak density which attend to over or under rain removal. Secondly, the results of single image after rain removal model often appear the phenomenon of variegated spots, image contrast saturation change and even unsmooth rain streak after rain removal. We use a three-way multi-scale weighted fusion module to enhance the feature extraction, and then generate an attention map through the improved spatial attentive module to accurately locate the location of the rain streaks. After the combination of the two, we will obtain the foreground information of the rain streaks. Through the characteristic of mutual game in the training mechanism of GAN, we can enhance the rain streak location recognition and effectively remove the rain at the same time. Through the training mechanism of the GAN network game, we can enhance the rain line location recognition and effectively remove the rain at the same time. Experiments show that our network achieves superior performance, it has high generalization for different rain streak density, and ensures that the contrast and saturation of the image are not changed.