Visual Text Region Detection in Video Streams
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing text extraction techniques from video frames are inefficient due to sensitivity to divergence and high computational cost, and they often focus on character detection rather than background noise removal, leading to increased memory requirements.
Innovation Solution
A method and system for extracting text from a video stream by determining visual text regions based on analysis of lines and curves, blobs, distribution along major axes, and common attributes, followed by segmentation from the background and inpainting with a predefined color.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If hierarchical clustering algorithms are used for text segmentation, then text can be segmented, but computational cost increases and sensitivity to divergence increases
Solution Approach 1:
The patent changes the parameter used for text segmentation from hierarchical clustering (which is sensitive to divergence and computationally expensive) to k-means clustering with a fixed number of clusters (k=2). This parameter change resolves the contradiction by using a simpler, more efficient algorithm that is less sensitive to initialization and convergence issues while maintaining segmentation accuracy.
Solution Approach 2:
The patent employs a simpler k-means clustering algorithm that can be easily discarded or replaced if needed, rather than relying on complex hierarchical clustering. The k-means approach uses a fixed number of iterations and simpler computational steps, making it more efficient and less prone to convergence problems.
2Measurement precision
If existing text detection techniques focus on character detection, then text characters can be detected, but background noise removal is insufficient and memory requirements increase
Solution Approach 1:
The patent segments the image into multiple regions based on color and texture characteristics before applying text detection. By dividing the image into distinct regions (e.g., foreground text regions vs. background regions), the system reduces the amount of data that needs to be processed and stored, thereby reducing memory requirements while maintaining detection accuracy.
Solution Approach 2:
The patent extracts and removes background noise by using color thresholding and region-based filtering. By taking out the background elements that are not text characters, the system reduces the data volume that needs to be stored and processed, thereby reducing memory requirements while improving text detection precision.
Data Source
AI summary
A method and system for extracting text from a video stream is disclosed. The method may include determining at least one visual text region in each image of a plurality of images. The at least one visual text region may include a plurality of text characters and determining the at least one visual text region is based on analysis of one of lines and curves associated with each of the plurality of text characters, blob associated with the plurality of text characters along an axis, distribution of the plurality of text characters along a major axis, and a common attribute associated with the plurality of text characters. The method may further include segmenting the plurality of text characters from a background, and inpainting each of the plurality of text characters with a predefined color.