Video Text Region Preprocessing for OCR Accuracy
Solution Overview
Problem
Current methods and systems are inefficient in preprocessing video regions containing text for improving optical character recognition and subsequent searching, failing to effectively utilize this information for retrieving related data.
Innovation Solution
A method and system that preprocesses video regions by locating and enhancing text images, binarizing them, segmenting touching characters, and identifying key words for improved optical character recognition and searching, using image processing techniques such as finite impulse response filters, adaptive thresholding, and discrete cosine transforms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing methods are used to conduct text-based searching, then searching capability is provided, but the determination of video regions containing text is inefficient and OCR performance is not improved
Solution Approach 1:
The patent applies preliminary action by preprocessing video frames before OCR processing. It detects potential text regions using motion detection and temporal consistency analysis, then applies preprocessing operations (binarization, filtering, enhancement) to these regions before OCR. This preliminary preparation improves OCR accuracy while reducing the computational burden compared to processing entire video frames.
Solution Approach 2:
The patent segments video content by identifying and isolating potential text regions from the rest of the video frame. It uses motion detection to find regions of interest, then further segments these regions through preprocessing steps that separate text from background. This segmentation allows focused processing on text-containing areas, improving both OCR performance and processing efficiency.
2Loss of information
If manual searching is conducted to obtain related information, then information can be retrieved, but user time and effort are significantly consumed
Solution Approach 1:
The patent implements self-service by enabling the system to automatically extract text from video content, perform OCR on the extracted text regions, and use the recognized text for searching. The system serves itself by automatically identifying text regions, preprocessing them, performing OCR, and using the results for information retrieval without requiring manual user intervention in the text extraction and search preparation process.
Solution Approach 2:
The patent replaces the mechanical manual searching process with an automated system that uses motion detection, temporal analysis, and OCR technology. Instead of users manually searching for information, the system automatically detects text regions in video content, extracts and recognizes the text, and performs searches based on the recognized text, significantly reducing user time and effort.
3Measurement precision
If text regions in video are not preprocessed, then video content is preserved in original form, but OCR input quality is insufficient for accurate recognition
Solution Approach 1:
The patent applies local quality by performing preprocessing operations only on identified text regions rather than the entire video frame. It uses motion detection and temporal consistency to locate potential text areas, then applies binarization, filtering, and enhancement operations specifically to these localized regions. This approach improves OCR input quality while minimizing the complexity and computational cost of preprocessing.
Solution Approach 2:
The patent changes parameters of the text region images to improve OCR input quality. It applies binarization to convert grayscale images to binary form, uses filtering to enhance text contrast and remove noise, and applies enhancement operations to sharpen text features. These parameter changes transform the raw video frames into optimized input suitable for accurate OCR recognition.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method and system for preprocessing text containing region of a video The invention provides a method and system for preprocessing the text containing region of video for improving the optical character recognition input.