In view of the importance of text information in video sequences and video indexes, this paper proposes a text localization algorithm based on text mixed features. The algorithm first performs edge detection and projection processing on single-frame images every 25 frames in the video sequence to extract text blocks, then uses the support vector basis to filter and eliminate the interference of non-text blocks, and finally uses the correlation between adjacent frames in the video sequence to search for text blocks in the remaining frames. The algorithm in this paper improves the detection speed while ensuring high detection accuracy.
You Might Like
Recommended ContentMore
Open source project More
Popular Components
Searched by Users
Just Take a LookMore
Trending Downloads
Trending ArticlesMore