Scholarlink

A new mipmap solution, which utilizes the Inverse Difference Pyramid (IDP) decomposition, is proposed in this work. Unlike the well-known pyramidal structures based on the Karhunen–Loeve transforms, IDP offers very high flexibility and efficiency in the storage and representation of various structures and forms, which could easily be implemented. Some of the most important advantages are the possibility to stop the decomposition in any level, where the required quality is obtained, and to use RSTC-invariant descriptions. Together with this, Deep learning (DL)-based solutions are available, which additionally accelerate the process of parameters selection optimization. The specific features of the new approach are presented theoretically, together with some experimental results. The computational complexity is relatively low, which opens wide application areas.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Mip-Map Structure Solution Based on the Adaptive Inverse Difference Pyramid Decomposition with Deep Learning Parameters Setting

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Lakhmi C. Jain,
  • Roumiana A. Kountcheva

摘要

A new mipmap solution, which utilizes the Inverse Difference Pyramid (IDP) decomposition, is proposed in this work. Unlike the well-known pyramidal structures based on the Karhunen–Loeve transforms, IDP offers very high flexibility and efficiency in the storage and representation of various structures and forms, which could easily be implemented. Some of the most important advantages are the possibility to stop the decomposition in any level, where the required quality is obtained, and to use RSTC-invariant descriptions. Together with this, Deep learning (DL)-based solutions are available, which additionally accelerate the process of parameters selection optimization. The specific features of the new approach are presented theoretically, together with some experimental results. The computational complexity is relatively low, which opens wide application areas.