Learning Adaptive Parameter Tuning for Image Processing

Abstract

The non-stationary nature of image characteristics calls for adaptive processing, based on the local image content. We propose a simple and flexible method to learn local tuning of parameters in adaptive image processing: we extract simple local features from an image and learn the relation between these features and the optimal filtering parameters. Learning is performed by optimizing a user defined cost function (any image quality metric) on a training set. We apply our method to three classical problems (denoising, demosaicing and deblurring) and we show the effectiveness of the learned parameter modulation strategies. We also show that these strategies are consistent with theoretical results from the literature.

Efficient Parameter Estimation in Image Processing using a Multi-Agent Hysteretic Q-Learning Approach

“…Qingang et al [11] proposed a decoupled learning methodology that dynamically fits the weights of a deep network as most existing trained models rely on the configuration of a single parameter. Jinming et al [12] proposed a simple method for learning local parameter tuning in adaptive image processing by extracting local characteristics from an image and learning the relationship between them and the optimal filtering parameters, optimizing any metric that defines the image's quality.…”