The low-light image enhancement aims to improve the brightness of images captured in insufficient lighting conditions while recovering color and edge details. In response to the current problems of poor color and edge restoration in low-light image enhancement algorithms based on deep learning, such as excessive or insufficient exposure and dark areas, this paper proposes a two-stage low-light image enhancement network (BMCENet) that focuses on brightness recovery curve estimation, multi-scale color correction, and edge refinement. The first stage introduces a Brightness Curve Estimation Block (BCE-Block) that iteratively maps curves to restore illumination for all pixels in the RGB channels. The second stage designs a multi-scale feature adjustment network guided by color and edge information, which includes a Color Feature Extraction Block (CFE-Block) and an Edge Feature Extraction Block (EFE-Block) designed to enhance color and edge details through learning features at different scales. The experimental results show that the objective metrics and subjective visual quality of the BMCENet proposed in this paper have achieved better performance on the mainstream datasets LOL-v1 and LOL-v2-syn.