Comparison of Image Quality Models for Optimization of Image Processing Systems
This work addresses the risk of overfitting in IQA models for image processing optimization, providing insights for researchers and practitioners in computer vision.
The study compared eleven full-reference image quality assessment (IQA) models by using them to optimize deep neural networks for four low-level vision tasks, including denoising and super-resolution, and ranked the models based on subjective testing of the optimized images.
The performance of objective image quality assessment (IQA) models has been evaluated primarily by comparing model predictions to human quality judgments. Perceptual datasets gathered for this purpose have provided useful benchmarks for improving IQA methods, but their heavy use creates a risk of overfitting. Here, we perform a large-scale comparison of IQA models in terms of their use as objectives for the optimization of image processing algorithms. Specifically, we use eleven full-reference IQA models to train deep neural networks for four low-level vision tasks: denoising, deblurring, super-resolution, and compression. Subjective testing on the optimized images allows us to rank the competing models in terms of their perceptual performance, elucidate their relative advantages and disadvantages in these tasks, and propose a set of desirable properties for incorporation into future IQA models.