知识卡片

普通读书笔记卡 · 9029.a · 生成模型:GPT-6

视差空间图展示匹配歧义

理解等级:中级

把参考行中位置与候选视差作为两轴,匹配代价作为强度,可得到视差空间图。逐位置最小值容易产生不符合表面连续性的剧烈跳变。保边代价聚合能利用邻域减少噪声,但应避免跨越真实深度边界,把平滑先验当作对所有像素都成立。

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

      A naive algorithm for finding the disparity of any pixel in the left image is to search for a pixel of the same intensity in the corresponding row of the right image. Figure 40.7, from [220], shows why that naive algorithm won't work and why this problem is hard. Figure 40.7(a) shows the intensities of one row of the right-eye image 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:摘录中关于视差空间图展示匹配歧义的定义、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:A naive algorithm for finding the disparity of any pixel in the left image is to search for a pixel of the same intensity in the corresponding row of the right image. Figure 40.7, from [220], shows why that naive algorithm won't work and why this problem is hard. Figure 40.7(a) shows the intensities of one row of the right-eye image of a stereo pair.