Face Image Quality Assessment Using CNN and Key Point Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In facial recognition systems, low-quality images can lead to recognition errors or system crashes, highlighting the need for effective image quality determination to improve efficiency and accuracy.
Innovation Solution
A method and apparatus that utilize a pre-trained convolutional neural network and key face point positioning model to assess the quality of face images by determining pixel categorization and key face point coordinates, calculating the probability of image obscuration, and evaluating image quality based on these probabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If facial recognition is performed directly on low-quality images, then processing speed is maintained, but recognition accuracy deteriorates and system stability worsens
Solution Approach 1:
The patent applies preliminary action by performing image quality assessment before facial recognition processing. The system evaluates quality metrics such as illumination, pose, and occlusion in advance, and only proceeds with recognition when quality thresholds are met, preventing wasted processing time on unsuitable images while ensuring recognition accuracy when performed.
Solution Approach 2:
The patent segments the facial recognition system into two distinct stages: image quality assessment and facial recognition processing. This segmentation allows independent optimization of each stage, with quality assessment filtering images beforehand and recognition processing only high-quality inputs, thereby improving both reliability and time efficiency.
2Reliability
If comprehensive image quality assessment is performed, then recognition accuracy is improved, but system complexity increases
Solution Approach 1:
The patent segments comprehensive quality assessment into multiple independent evaluation dimensions including illumination assessment, pose assessment, and occlusion assessment. Each dimension is evaluated separately using dedicated algorithms, and results are combined to determine overall image quality. This segmentation makes the complex assessment process more manageable and implementable.
Solution Approach 2:
The patent applies local quality assessment by evaluating specific regions and attributes of the image separately. Instead of treating the image as a whole, the system assesses local characteristics such as lighting conditions in different face regions, pose angles of specific facial features, and occlusion levels of individual facial components, enabling comprehensive quality evaluation through localized analysis.
3Productivity
If image quality assessment is performed before recognition, then recognition efficiency is improved, but processing overhead increases
Solution Approach 1:
The patent applies partial action by performing selective quality assessment focusing on the most critical quality factors for facial recognition success. Rather than conducting exhaustive analysis of all possible image attributes, the system prioritizes assessment of key parameters such as illumination adequacy, pose acceptability, and occlusion levels, achieving sufficient quality evaluation with reduced processing overhead.
Data Source
AI summary
Embodiments of the present disclosure disclose a method and apparatus for determining image quality. The method comprises: acquiring a to-be-recognized image and facial region information used for indicating a facial region in the to-be-recognized image; extracting a face image from the to-be-recognized image on the basis of the facial region information; inputting the face image into a pre-trained convolutional neural network to obtain probabilities of each pixel comprised in the face image belonging to a category indicated by each category identifier in a preset category identifier set; inputting the face image into a pre-trained key face point positioning model to obtain coordinates of each key face point comprised in the face image; determining a probability of the face image being obscured on the basis of the probabilities and the coordinates; and determining whether the quality of the face image is up to standard on the basis of the probability.