期刊文献+

基于改进单深层神经网络的自然场景中维吾尔文检测 被引量:1

Uyghur text detection in natural scene based on improved single deep neural network
下载PDF
导出
摘要 针对自然场景中维吾尔文检测难度大的问题,改进单深层神经网络对自然场景中维吾尔文进行检测。该网络结构由维吾尔文特征提取组件和多层特征融合的文本检测组件组成,以端到端的方式训练学习预测维吾尔文文本框的位置以及置信度。维吾尔文特征提取组件利用卷积神经网络提取自然场景维吾尔文图像中的多尺度和多层级维吾尔文特征。多层特征融合的文本检测组件则使用维吾尔文特征提取组件提取的特征,预测文本框的位置和维吾尔文类别的置信度。分析发现与中英文检测不同,维吾尔文文本具有更特殊的特征,针对这种特性设计了多宽高比和多尺寸大小的默认框并调整了部分卷积核的大小。经在自然场景中具有维吾尔文的图片集实验表明,改进的单深层神经网络方法考虑了图像的多尺度和多层级特征对检测精度的影响,算法的准确率和F值分别达到了0. 723 4和0. 611 5,提高了检测的准确率。 In order to overcome the difficulties of detecting the Uyghur text in natural scene,this paper improved a single deep neural network to detect Uyghur text in natural scene. The network structure combined the Uyghur feature extraction and the multi-layer features fusion text detection component. What was more,it predicted the position of Uyghur text bounding box and the confidence score of Uyghur text in an end-to-end manner. Uyghur feature extraction component used convolutional neural network to extract multi-scale and multi-level Uyghur features from natural Uyghur images. The multi-layer features fusion text detection component made use of the features extracted by the Uyghur feature extraction component to predict the position of the Uyghur text bounding boxes and the confidence of the Uyghur category. The analysis shows that Uyghur text has more special features than English and Chinese texts. For this feature,it designed a default box with multiple aspect ratios and multiple sizes and adjusted the size of some convolution kernels. Experiments on Uyghur natural scene images show that the improved single deep neural network method considers the influence of multi-scale and multi-level image features on the detection accuracy and improves the detection accuracy. The accuracy and the F value of the algorithm respectively reach 0. 723 4 and 0. 611 5.
作者 彭勇 哈力旦·阿布都热依木 丁维超 Peng Yong;Halidan·Abudureyimu;Ding Weichao(School of Electrical Engineering,Xinjiang University,Urumchi 830049,China;Suzhou Research Institute,Southeast University,Suzhou Jiangsu 215123,China)
出处 《计算机应用研究》 CSCD 北大核心 2019年第9期2876-2880,共5页 Application Research of Computers
基金 新疆维吾尔自治区自然科学基金资助项目(2016D01C048)
关键词 维吾尔文检测 单深层神经网络 多尺度特征 Uyghur text detection single deep neural network multi-scale features
  • 相关文献

参考文献3

二级参考文献15

共引文献16

同被引文献7

引证文献1

二级引证文献2

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部