A New Vision For Artistic Segmentation

Abstract: In this project, we motivate the need for a new class of segmentation algorithms that partition images in ways that are artistically driven and aesthetically pleasing. Inspired by past work in computational parquetry, we present a novel algorithm to create a spatial feature-driven image segmentation that appeals to certain artistic forms and complements textural feature-based segmentation and matching. We hope that this contribution will motivate other researchers to explore how computer vision techniques can be leveraged to create art that transcends the digital into the physical. Paper