search

cs-chan / Total-Text-Dataset

Total Text Dataset. It consists of 1555 images with more than 3 different text orientations: Horizontal, Multi-Oriented, and Curved, one of a kind.
BSD 3-Clause "New" or "Revised" License
747 stars 140 forks source link

How to bound text region with polygons after connected component analysis #5

Closed xieenze closed 6 years ago

xieenze commented 6 years ago

Hello, I am confused about some details. In your paper you first do segmentation on image, then binarize the saliency map and do connected component analysis. However, I don't know how to bound them tightly with polygons? Can you share some details?

ckchng commented 6 years ago

Sure. First of all, you get pixel map as the output from the Deconvnet. 1) We use the Matlab function 'bwconncomp' to get the CC. 2) Then use the Matlab function 'regionprops' to retrieve the 'Convexhull', which is all the polygon vertices of every CCs.