← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9030.a · 生成模型:GPT-6

方向直方图描述局部结构

理解等级:中级

SIFT以局部梯度方向与空间分块直方图描述关键点邻域,减少轻微光照和位置变化对匹配的影响。检测、尺度和方向归一化与描述子是不同环节;它对某些变化稳健,并非任意光照、视角或非刚性变形下都不变。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      Matching corresponding image features require a local description of the image region to allow matching like regions across two images. SIFT [309] uses histograms of oriented filter responses to create an image descriptor that is sensitive to local image structure, but that is insensitive to minor changes in lighting or location which may occur across the two images of a stereo pair.

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第40章《立体视觉》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)/_epub-src/OPS/xhtml/c040.xhtml) - 结论依据:摘录中关于方向直方图描述局部结构的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:Matching corresponding image features require a local description of the image region to allow matching like regions across two images. SIFT [309] uses histograms of oriented filter responses to create an image descriptor that is sensitive to local image structure, but that is insensitive to minor changes in lighting or location which may occur across the two images of a stereo pair.