Scholarlink

Abstract <p>The problem on color space transform as a stage of the digital image formation pipeline is investigated. The available methods for solving this problem mainly perform the pointwise image transformation without considering the contextual information. The proposed new color space transform method based on the LW-ISP convolutional neural network can take into account the scene semantics. The proposed method demonstrates significant advantages over the modern color transform approaches, specifically, a 39% reduction of the CIEDE2000 color difference and a 35% reduction of the angular distance in the ProLab space.</p>

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

Convolutional Neural Networks in the Color Space Transform Problem

  • 期刊论文

  • 发表日期：

    2025年8月1日

查看全文
  • A. S. Sribniak,
  • A. V. Panshin,
  • E. I. Ershov

摘要

Abstract

The problem on color space transform as a stage of the digital image formation pipeline is investigated. The available methods for solving this problem mainly perform the pointwise image transformation without considering the contextual information. The proposed new color space transform method based on the LW-ISP convolutional neural network can take into account the scene semantics. The proposed method demonstrates significant advantages over the modern color transform approaches, specifically, a 39% reduction of the CIEDE2000 color difference and a 35% reduction of the angular distance in the ProLab space.