Search

Recognizing texts in images plays an important role in many applications, such as industrial intelligence, robot vision, automatic driving, command assistance, and scene understanding. Although great progress has been achieved in various fields, research on complex systems modeling using text recognition technology requires further attention. To address this, we propose a new end-to-end multi-task learning method, which includes a super-resolution branch (SRB) and a recognition branch. To effectively learn the semantic information of images, we utilize the feature pyramid network (FPN) to fuse high- and low-level semantic information. The feature map generated by FPN is then delivered separately to the super-resolution branch and the recognition branch. We introduce a novel super-resolution branch, the SRB based on the proposed dual attention mechanism (DAM), designed to enhance the capability of learning low-resolution text features. The DAM incorporates the residual channel attention to enhance channel dependencies and the character attention module to focus on context information. For the recognition branch, the feature map generated by FPN is fed into an RNN sequence module, and an attention-based decoder is constructed to predict the results. To address the issue of low-resolution text recognition in numerous Chinese scenes, we propose the Chinese super-resolution datasets instead of relying on traditional down-sampling techniques to generate training datasets. Experiments demonstrate that the proposed method performs robustly on low-resolution text images and achieves competitive results on benchmark datasets. [ABSTRACT FROM AUTHOR]

A study is presented of the geographical distribution and spatial and temporal variabilities of the western China snow cover in the past 47 yr between 1951 and 1997. The data used consist of Scanning Multichannel Microwave Radiometer (SMMR) 6-day snow-depth charts, NOAA weekly snow extent charts, and the daily snow depth and number of snow cover days from 106 selected meteorological stations across western China. Empirical orthogonal function was performed on the SMMR dataset to better understand the spatial pattern and variability of the Qinghai–Xizang (Tibet) snow cover. A multiple linear regression analysis was conducted to show the association of interannual variations between snow cover and snow season temperature as well as precipitation. Further, the autoregressive moving average model was fitted to the snow and climate time series to test for their long-term trends. Results show that western China did not experience a continual decrease in snow cover during the great warming period of the 1980s and 1990s. It is of interest to note that no correlation was identified between temperature and precipitation in the snow cover season. However, year-to-year fluctuation of snow cover responds to both snowfall and snow season temperature. About one-half to two-thirds of the total variance in snow cover is explained by the linear variations of snowfall and snow season temperature. The long-term variability of western China snow cover is characterized by a large interannual variation superimposed on a small increase trend. The positive trend of the western China snow cover is consistent with increasing snowfall, but is in contradiction to regional warming. In addition, many constraints of the Qinghai–Xizang (Tibet) snow cover force the author’s challenge of Blanford’s hypothesis. [ABSTRACT FROM AUTHOR]