Self-adaptive Single and Multi-illuminant Estimation Framework based on Deep Learning
This work addresses a key issue in computer vision for camera systems, offering a practical solution for more accurate color correction in varied lighting conditions, though it is incremental with novel method components.
The paper tackles the problem of illuminant estimation in digital camera pipelines, which reduces color casting from non-white illuminants, by proposing a self-adaptive framework that handles both single and multi-light-source scenes, achieving a 16% improvement over state-of-the-art methods on standard benchmarks.
Illuminant estimation plays a key role in digital camera pipeline system, it aims at reducing color casting effect due to the influence of non-white illuminant. Recent researches handle this task by using Convolution Neural Network (CNN) as a mapping function from input image to a single illumination vector. However, global mapping approaches are difficult to deal with scenes under multi-light-sources. In this paper, we proposed a self-adaptive single and multi-illuminant estimation framework, which includes the following novelties: (1) Learning local self-adaptive kernels from the entire image for illuminant estimation with encoder-decoder CNN structure; (2) Providing confidence measurement for the prediction; (3) Clustering-based iterative fitting for computing single and multi-illumination vectors. The proposed global-to-local aggregation is able to predict multi-illuminant regionally by utilizing global information instead of training in patches, as well as brings significant improvement for single illuminant estimation. We outperform the state-of-the-art methods on standard benchmarks with the largest relative improvement of 16%. In addition, we collect a dataset contains over 13k images for illuminant estimation and evaluation. The code and dataset is available on https://github.com/LiamLYJ/KPF_WB