Final Project for CS283: Advanced Computer Vision
Texture synthesis is a significant application of image pyramids, leveraging multi-resolution representations for analyzing images across various spatial scales. Steerable pyramids, a type of image pyramid, offer numerous advantages, providing a rich representation that includes information about image scale and orientation. In this tutorial, we explore the implementation of the Heeger-Bergen algorithm for synthesizing grayscale textures using steerable pyramids. The tutorial extends the algorithm to adapt to different image sizes, explores color textures, and addresses the handling of edges in textures.
The packages required for running the tutorial can be found in requirements.txt. Install using pip install -r requirements.txt.
The file HB_Texture_Synthesis_Problem_Set_Problems.ipynb contains the problem set, and HB_Texture_Synthesis_Problem_Set_Solution.ipynb contains solutions to the problem set.