← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9166.d · 生成模型:GPT-6

描述子归一化只抵抗部分光变

理解等级:中级

归一化、限制强分量后再归一化,可减轻整体对比度和某些强梯度变化的影响,但遮挡、饱和、非均匀光照和大视角变化仍可能破坏匹配。所谓不变性是在建模与近似条件下成立,不能理解为任何拍摄变化都相同。

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

      7. Normalize, threshold, and again normalize the feature descriptor vector to make it more robust against illumination changes. 8. Group local gradient magnitude and orientation into bins and create a 128-dimensional feature descriptor. The resulting 128-dimensional feature vectors are now scale-invariant (because of step 2), rotation-invariant (result of step 5), and robust to illumination changes (result of step 7). 9.4.2 Object Recognition Using Scale-Invariant Features

    参考来源

    - 位置:《Introduction to Autonomous Robots Mechanisms, Sensors, Actuators, and Algorithms (Nikolaus Correll, Bradley Hayes etc.)》第09章《特征提取》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Introduction to Autonomous Robots Mechanisms, Sensors, Actuators, and Algorithms (Nikolaus Correll, Bradley Hayes etc.)/_epub-src/OEBPS/xhtml/chapter_9.xhtml) - 结论依据:摘录中关于描述子归一化只抵抗部分光变的定义、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:7. Normalize, threshold, and again normalize the feature descriptor vector to make it more robust against illumination changes. 8. Group local gradient magnitude and orientation into bins and create a 128-dimensional feature descriptor. The resulting 128-dimensional feature vectors are now scale-invariant (because of step 2), rotation-invariant (result of step 5), and robust to illumination changes (result of step 7). 9.4.2 Object Recognition Using Scale-Invariant Features