No-Reference Image Quality Assessment Based on Dual-Domain Feature Fusion

Area of Science:

• Computer Vision
• Image Processing
• Signal Processing

Background:

• Image quality assessment (IQA) is crucial for evaluating digital images perceptually.
• Existing no-reference IQA models often lack comprehensive feature extraction.
• Computational models aim to mimic human visual system (HVS) judgments of image quality.

Purpose of the Study:

• To propose a novel no-reference image quality assessment (IQA) model named DFF-IQA.
• To enhance the accuracy and perceptual consistency of computational IQA.
• To develop a robust method for evaluating image quality without a reference image.

Main Methods:

• Feature extraction in both spatial (weighted local binary pattern, naturalness, spatial entropy) and frequency domains (spectral entropy, oriented energy distribution).
• Fusion of dual-domain features to create a quality-aware feature vector.
• Quality regression using a random forest model to predict image quality scores.

Main Results:

• The DFF-IQA model demonstrated superior performance on the LIVE image quality database.
• Experimental results indicate higher consistency with human visual perception compared to existing IQA models.
• The dual-domain feature fusion approach effectively captures essential image quality attributes.

Conclusions:

• The proposed DFF-IQA model offers a significant advancement in no-reference image quality assessment.
• Dual-domain feature fusion is an effective strategy for developing perceptually relevant IQA models.
• The DFF-IQA method provides a reliable computational tool for evaluating image quality.