Computer Vision Foundations · 免费预览
宽针孔与边缘相机
平面国中的两种相机（其成像矩阵会模糊和积分），以及它们的逆对噪声的作用。
1. Widening the pinhole
Widening the pinhole
2. The edge camera integrates
The edge camera integrates
3. Why the inverse of an integrator is dangerous
Why the inverse of an integrator is dangerous
4. The regularized inverse smooths instead
The regularized inverse smooths instead
5. Reading the edge camera's row sum
A flatland edge camera with five sensor elements sees a world row of five samples. The first sensor element gathers light from every world sample while the last gathers light from only one. Why do the rows of the imaging matrix grow shorter as you move down the matrix?
6. What the exact inverse does with noise
An edge camera's readings contain a single noisy reading that is too large by a small amount. The scene is recovered with the exact inverse of the imaging matrix. What does the recovered scene look like?
7. The cumulative sum reading
A flatland edge camera sees the world row and the first sensor element sums everything to its right, which is the entire row. Compute the first reading and write the number.