Edge detail enhancement method of low-illumination image based on deep self-encoding network
Traditional enhancement methods often rely on manually designed feature extractors or simple filtering techniques,but their effectiveness is limited when processing low light images,and they cannot fully preserve image details,resulting in a de-crease in image quality,manifested as blurred edges,roughness,and lack of clarity.To address this issue,this study proposes a low illumination image edge detail enhancement method based on deep autoencoder networks.This method locates detail loss areas by setting neighborhood edge matrices,extracts and smooths the boundaries of these areas,and achieves the reconstruction of low illu-mination image edges.Using deep autoencoder networks to extract key edge detail features layer by layer abstractly,and further pro-cessing these features using variable density enhancement to optimize and enhance the edge details of low illumination images.Ex-perimental results have shown that this method can obtain higher edge pixels and improve image resolution,with practical applica-tion potential and value.
deep self coding networklow light imagesedge detailsenhancement methods
万方数据
