RWYI: Reading What You Are Interested in with a Learning-Based Text Interactive System

Algorithm 1

The overall process of the system RWYI.

	Model: HTD, FCSRN, CRNN
	Input: Image or video frame in text reading scenarios .
	Output: The content of the text of interest to the reader .
	WhileDo
	Stage 1:
	Hand-text hybrid object detection
	If there is a hybrid hand-text object Then
	Obtain hand-text hybrid object prediction boxes
	Screening to get the best prediction box
	Low-quality image of the text area
	Else
	End
	Stage 2:
	Text image superresolution processing
	IfThen
	Aligned low-quality text image
	High-quality text area image , font-consistency prior
	Else
	End
	Stage 3:
	Text recognition
	If Then
	The content of the text of interest
	Else
	End
	If Then
	Continue to the next image or video frame
	return
	End
	End