MINER

Intro

Unofficial Pytorch-lightning implement of MINER: Multiscale Implicit Neural Representations(Siggraph2022)

How it work?

What's Gaussian pyramids?

In a Gaussian pyramid, subsequent images are weighted down using a Gaussian average (Gaussian blur) and scaled down. Each pixel containing a local average corresponds to a neighborhood pixel on a lower level of the pyramid. This technique is used especially in texture synthesis.

What's Laplacian pyramids?

A Laplacian pyramid is very similar to a Gaussian pyramid but saves the difference image of the blurred versions between each levels. Only the smallest level is not a difference image to enable reconstruction of the high resolution image using the difference images on higher levels. This technique can be used in image compression.

Given an input image of $(2^N+1) \times (2^N+1)$, the Laplacian pyramid representation consists of :

• $L_0, ...,L_{k-1}$
• $g_k$ for some $K\le N$

Moreover

• $L_l = g_l - \text{EXPAND}(g_{l+1})$
• $\text{EXPAND}(g_l) = 4\sum^2_{m=-2}\sum^2_{n=-2}w(m,n)g_l(\frac{i-m}{2},\frac{j-n}{2} )$

MINER Algorithm

How to setup?

Data preparation

• build a directory called "data/"
• make sure your own images put in "data/"
• Data used in My Experiment:
  • NASA's Pluto image: NASA
  • My rendering artwork: geo

Acknowledgement

Thanks for all the contributors below!

kwea123's repo: https://github.com/kwea123/MINER_pl

kwea123's videos: YouTube

Wikipedia: Pyramid (image processing)

University of Toronto CSC320 slides: https://www.cs.toronto.edu/~mangas/teaching/320/slides/CSC320L10.pdf