Multi-oriented text detection from natural scene images based on a CNN and pruning non-adjacent graph edges
作者：
Highlights：
•
• A novel method for multi-oriented scene text detection is presented.
• A CNN classifier is used to eliminate non-character candidates.
• We argue that the multi-oriented text line construction problem can be posed as one of splitting an undirected, fully connected graph into connected sub-graphs.
• An effective coarse-to-fine algorithm is proposed to prune the non-adjacent edges, those in which two corresponding characters are not adjacent to each other in the same word.
• The presented method is robust to the multi-oriented scene text detection.
摘要
•A novel method for multi-oriented scene text detection is presented.•A CNN classifier is used to eliminate non-character candidates.•We argue that the multi-oriented text line construction problem can be posed as one of splitting an undirected, fully connected graph into connected sub-graphs.•An effective coarse-to-fine algorithm is proposed to prune the non-adjacent edges, those in which two corresponding characters are not adjacent to each other in the same word.•The presented method is robust to the multi-oriented scene text detection.
论文关键词：Text detection,Scene image,Multi-orientation,CNN
论文评审过程：Received 29 September 2017, Revised 28 February 2018, Accepted 28 February 2018, Available online 8 March 2018, Version of Record 22 March 2018.
论文官网地址：https://doi.org/10.1016/j.image.2018.02.016
