Perception-Distortion Balanced Super-Resolution: A Multi-Objective Optimization Perspective
This work addresses a key challenge in image restoration for applications like photography and medical imaging, offering an incremental improvement over existing methods by optimizing conflicting objectives more effectively.
The paper tackles the perception-distortion trade-off in image super-resolution by formulating it as a multi-objective optimization problem, combining gradient-free evolutionary algorithms with gradient-based Adam to train models that achieve better perceptual quality and reconstruction fidelity than competitors, with experiments showing improved results using the same backbone network.
High perceptual quality and low distortion degree are two important goals in image restoration tasks such as super-resolution (SR). Most of the existing SR methods aim to achieve these goals by minimizing the corresponding yet conflicting losses, such as the $\ell_1$ loss and the adversarial loss. Unfortunately, the commonly used gradient-based optimizers, such as Adam, are hard to balance these objectives due to the opposite gradient decent directions of the contradictory losses. In this paper, we formulate the perception-distortion trade-off in SR as a multi-objective optimization problem and develop a new optimizer by integrating the gradient-free evolutionary algorithm (EA) with gradient-based Adam, where EA and Adam focus on the divergence and convergence of the optimization directions respectively. As a result, a population of optimal models with different perception-distortion preferences is obtained. We then design a fusion network to merge these models into a single stronger one for an effective perception-distortion trade-off. Experiments demonstrate that with the same backbone network, the perception-distortion balanced SR model trained by our method can achieve better perceptual quality than its competitors while attaining better reconstruction fidelity. Codes and models can be found at https://github.com/csslc/EA-Adam}{https://github.com/csslc/EA-Adam.