Lightweight Curve Estimation Method for Low-Light Image Enhancement

Low-light image enhancement is an important issue in digital image processing area. However, in practice, it is still a difficult problem, especially how to balance computational cost and restore result. To tackle this difficulty, in this paper, we present a fast curve transition based low-light image enhancement model and propose a new lightweight curve estimation model. Specifically, an extremely lightweight curve estimation network based on parameter curve model is proposed, called LCE. Compared with the traditional method, called Zero-DCE, which utilizes iterative polynomial interpolation, the proposed can estimate the translate curve non-recursively. Benefitting from intuitive and simple nonlinear curve mapping as well as the lightweight curve estimation network, experimental results show that the proposed model is superior to the existing methods in terms of computational cost, while holds similar enhanced results on low-light images.