Laplacian Pyramid 拉普拉斯金字塔

Gaussian Pyramid vs Laplacian Pyramid

A Laplacian Pyramid is a linear invertible image representation consisting of a set of band-pass images, spaced an octave apart, plus a low-frequency residual. Formally, let $d(\cdot )$ be a downsampling operation which blurs and decimates a $j\times j$ image $I$, so that $d(I)$ is a new image of size $\frac{j}{2}\times \frac{j}{2}$. Also, let $u(\cdot )$ be an upsampling operator which smooths and expand $I$ to be twice the size, so $u(I)$ is a new image of size $2j\times 2j$. We first build a Gaussian pyramid $G(I)=[I_0,I_1,...I_K],$ where $I_0=I$ and $I_k$ is $k$ repeated applications of $d(\cdot )$ to $I$. $K$ is the number of levels in the pyramid, selected so that the final level has very small spatial extent ($\le 8\times 8$ pixels).

The coefficients $h_k$ at each level $k$ of the Laplacian pyramid $L(I)$ are constructed by taking the difference between adjacent levels in the Gaussian pyramid, upsampling the smaller one with $u(\cdot )$, so that the sizes are compatible:

$h_k=L_k(I)=G_k(I)-u(G_{k+1}(I))=I_k-u(I_{k+1})$

Intuitively, each level captures images structure present at a particular scale. The final level of the Laplacian pyramid $h_K$ is not a difference image, but a low-frequency residual equal to the final Gaussian pyramid level, i.e. $h_K=IK$. Reconstruction from a Laplacian pyramid coefficients $[h_1,...h_K]$ is performed using the backward recurrence:

$I_k=u(I_{k+1})+h_k$

which is started with $I_K=h_K$ and the reconstructed image being $I=I_o$. In other words, starting at the coarsest level, we repeatedly upsample and add the difference image h at the next finer level until we get back to the full resolution image.

补充

[OpenCV / The Laplacian pyramid] 浅析拉普拉斯金字塔 / 利用拉普拉斯完成采样的恢复 学习笔记

Reference

1. Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks
2. Laplacian Pyramid