Video Image Entropy Screening to Suppress False-Positive Detections
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Object detection algorithms based on artificial intelligence are prone to false-positive detections due to non-domain objects such as artifacts and blurred images, which distort actual information and increase computational load.
Innovation Solution
Suppression of potentially false-positive detections in video images by calculating and indicating information content using extended entropy metrics and similarity characteristics, and suppressing detections based on threshold values and known disturbance characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If object detection algorithms are applied to all images in a video signal, then detection coverage is improved, but computational load increases and false-positive detections increase
Solution Approach 1:
The patent applies preliminary action by calculating entropy metrics and disturbance characteristics before performing object detection. This pre-screening process identifies images that are likely to produce false positives (such as blurred images, artifacts, or images without relevant content), allowing the system to skip detection on these images and thus reduce computational load while maintaining detection coverage on meaningful images.
2Reliability
If object detection algorithms are applied to all images in a video signal, then detection coverage is improved, but false-positive detections increase
Solution Approach 1:
The patent calculates entropy metrics and disturbance characteristics before object detection to pre-identify images likely to produce false positives. By filtering out blurred images, artifacts, and images without relevant content beforehand, the system maintains high detection coverage on meaningful images while significantly reducing the false-positive rate.
Solution Approach 2:
The patent introduces entropy metrics and disturbance characteristics as intermediary variables between the raw video signal and the object detection algorithm. These intermediaries serve as a filtering layer that assesses image quality and relevance, allowing the system to make informed decisions about which images warrant detection processing, thereby reducing false positives while maintaining coverage.
3Reliability
If object detection is performed on images without information content, then detection completeness is improved, but computing power is wasted
Solution Approach 1:
The patent performs preliminary entropy calculation and disturbance characteristic analysis before object detection to identify images without information content. By detecting blurred images, artifacts, or images with low entropy early in the processing pipeline, the system can skip detection on these images, thereby maintaining detection completeness on meaningful images while eliminating wasted computing power on irrelevant images.
Data Source
AI summary
An information content of a section of a current image of a series of images of a video signal is calculated, wherein the video signal has to be fed to an algorithm for calculating and indicating detections of objects in the video signal. If the calculated information content of the section of the current image is smaller than a threshold value, the calculation and indication of detections of objects for the section of at least the current image or the current image and further images of the series of images is suppressed.