Video Automation OCR Using Timer-Based Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Typical OCR test functions in video or video on demand (VOD) test automation produce unpredictable and inconsistent results, fail to support dynamic comparison of text and data from multiple video frames, and lack control over execution time, leading to unacceptably low reliable results and often requiring human validation or manual testing.
Innovation Solution
A system that uses a timer-based correlation technique to recognize information in screen snapshots, allowing for the construction of a reference library and determining performance information for various video signals, including IP video and non-IP video, by capturing and processing multiple frames within a specified time duration using OCR-based recognition methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If indexed looping is used to run OCR multiple times, then the number of OCR comparisons increases, but the reliability of results decreases due to false matches and uncontrollable execution time
Solution Approach 1:
The system dynamically adjusts the number of frame comparisons and timing parameters based on video content characteristics rather than using fixed indexed loops. The correlation process adapts to varying video scenarios, selecting optimal comparison counts and time windows to maximize reliable matches while minimizing false positives.
Solution Approach 2:
The system changes multiple parameters simultaneously including timing thresholds, correlation confidence levels, and frame selection criteria to optimize OCR comparison reliability. By adjusting these parameters based on video characteristics, the system achieves consistent results without relying on arbitrary loop counts.
2Measurement precision
If typical OCR test functions are used in video automation, then text recognition is performed, but execution time becomes uncontrollable and results are unpredictable
Solution Approach 1:
The system performs preliminary analysis of video frames to identify optimal comparison windows and timing parameters before executing OCR comparisons. By pre-determining when and how many frames to compare based on video content analysis, the system controls execution time while maintaining recognition accuracy.
Solution Approach 2:
The system uses feedback from initial frame correlations to adjust subsequent OCR comparison parameters. Based on the quality and timing of detected text in early frames, the system dynamically modifies comparison thresholds and time windows to achieve consistent results within controlled execution time.
3Reliability
If manual testing is used instead of automated OCR, then result validation is more reliable, but productivity and efficiency decrease
Solution Approach 1:
The system performs self-validation through multiple correlated OCR comparisons across video frames, automatically verifying text recognition results without requiring manual intervention. The correlation process inherently validates results by checking consistency across multiple frames and comparison attempts, maintaining high reliability while preserving automation efficiency.
4Quantity of substance
If OCR comparisons are run multiple times with indexed looping, then more data points are collected, but false matches increase and reduce result reliability
Solution Approach 1:
The system dynamically determines the optimal number of comparisons based on video characteristics and correlation results, rather than using fixed indexed loops. The comparison process adapts to the specific video content, increasing comparisons only when necessary and stopping when sufficient confidence is achieved, thereby reducing false matches.
Solution Approach 2:
The system replaces mechanical indexed looping with an intelligent correlation-based selection process that uses video content analysis to determine comparison parameters. This substitution eliminates the arbitrary nature of fixed-loop counters and replaces them with adaptive, content-driven decision-making that reduces false matches.
Data Source
AI summary
A video analysis system may utilize display screen snapshots captured from a device. The system may receive time information that indicates a time duration, position-related information that indicates a region, and reference information that indicates information expected to appear in the region during the time duration. The system may transmit an instruction to capture display screen snapshots during a time duration indicated by the time information. The system may receive a display screen snapshot and recognize information in the region indicated by the position-related information. The system may then determine whether the recognized information corresponds to the reference information and based on the result determine whether the video test automation has passed.