摘要
提出了一种改进的稀疏表示图像重建算法,旨在提升旅游地标图像的重建质量,该算法通过自适应结构化字典学习、多约束优化函数和非局部相似性增强三大技术改进,克服了传统方法在特征冗余性、多尺度细节丢失和噪声敏感性方面的挑战。实验结果表明,本方法在重建精度上显著优于传统的插值法、深度学习方法和稀疏编码方法,尤其在细节复原和结构保持上表现出色。
An improved sparse representation image reconstruction algorithm is proposed to improve the reconstruction quality of tourist landmark images.The algorithm overcomes the challenges of traditional methods in feature redundancy,multi-scale detail loss and noise sensitivity through three technical improvements:adaptive structured dictionary learning,multi-constraint optimization function and non-local similarity enhancement.The experimental results show that the proposed method is significantly superior to the traditional interpolation method,deep learning method and sparse coding method in reconstruction accuracy,especially in detail restoration and structure preservation.
作者
王冠
来思渊
WANG Guan;LAI Siyuan(Zhejiang Vocational College of Tourism,Hangzhou 310000,China;Zhejiang Sci-Tech University,Hangzhou 310018,China)
出处
《佳木斯大学学报(自然科学版)》
2025年第7期140-143,共4页
Journal of Jiamusi University:Natural Science Edition
基金
2021年浙江省教育厅一般科研项目(Y202147712)
2022年度浙江省教育厅高校国内访问工程师“校企合作项目”(FG2022058)
2024年度浙江旅游职业学院校级专项重点课题(2024ZXZD02)。
关键词
稀疏表示
图像重建
超分辨率
旅游地标
sparse representation
image reconstruction
super resolution
tourist landmark
