Image Blur Detection Using Face Area Overlap Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users face challenges in managing large numbers of images taken with devices like digital cameras and mobile phones, as existing tools struggle to efficiently identify and remove undesirable images, particularly those with blur issues, without user intervention.
Innovation Solution
A method and apparatus that detect face and blur areas in images, calculate the overlap ratio between them, and determine if the image is abnormal based on this analysis, allowing for automatic correction or removal of abnormal blur images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If existing picture management tools are used to organize and identify undesirable images, then images can be organized by shooting time, tag, location, color, event, etc., but the tools cannot automatically distinguish between intended and unintended blur, requiring manual user intervention to review and remove undesirable images
Solution Approach 1:
The image is divided into multiple frequency bands using wavelet transform, allowing separate analysis of different frequency components. This segmentation enables the system to identify blur-specific frequency characteristics without analyzing the entire image spectrum, reducing computational complexity while maintaining automation.
Solution Approach 2:
A wavelet transform intermediary is introduced to bridge the gap between raw image data and blur detection. The wavelet transform converts the image into a frequency domain representation that highlights blur characteristics, enabling automatic detection without requiring complex direct analysis of the original image data.
2Productivity
If manual review of images is required to identify and remove blur images, then users can make informed decisions about which images to keep, but this process is time-consuming and reduces productivity
Solution Approach 1:
The system performs preliminary blur detection and classification automatically before user review. By pre-identifying images with unintended blur through frequency analysis, the system filters out obvious candidates, reducing the time users need to spend on manual review while maintaining high productivity.
Solution Approach 2:
The image management system performs self-service by automatically detecting and flagging undesirable blur images without requiring continuous user intervention. The wavelet-based blur detection algorithm operates autonomously to identify images needing review, freeing users from manual inspection of every image.
3Reliability
If all blur images are treated as undesirable and automatically removed, then image quality is improved, but intended artistic blur effects are also lost
Solution Approach 1:
The system applies different evaluation criteria to different regions and contexts within images. By analyzing local frequency characteristics and comparing them against patterns associated with unintended blur versus artistic blur, the system can distinguish between the two types, maintaining reliability while adapting to different blur scenarios.
Solution Approach 2:
The blur detection system is dynamic and adaptive, adjusting its classification based on image content and context. Rather than applying a static threshold to all images, the system dynamically evaluates frequency patterns to determine whether blur is unintended or artistic, enabling accurate detection while preserving versatility.
Data Source
AI summary
A method, apparatus, and computer readable recording medium for processing an image. The method includes detecting a face area and a blur area from an image; checking a degree of overlap between the face area and the blur area by comparing a location of the face area with a location of the blur area; and determining whether the image is an abnormal image according to the degree of overlap between the face area and the blur area.