Real-Time Pattern Matching Using Projection Kernels
IEEE Transactions on Pattern Analysis and Machine Intelligence
Synthesizing transition textures on succession patterns
GRAPHITE '05 Proceedings of the 3rd international conference on Computer graphics and interactive techniques in Australasia and South East Asia
GI '07 Proceedings of Graphics Interface 2007
Journal of Computer Science and Technology
Fast image replacement using multi-resolution approach
ACCV'06 Proceedings of the 7th Asian conference on Computer Vision - Volume Part II
Hi-index | 0.00 |
Patch-based texture synthesis has proven to produce highquality textures faster than pixel-based approaches. Previousalgorithms differ in how the regions of overlap betweenneighboring patches are treated. We present an approachthat produces higher quality overlap regions than simpleblending of patches or computing good boundaries, however,that is faster than re-synthesizing invalid pixels using aclassical per-pixel synthesis algorithm: we use a k-nearestneighbor (knn) data structure, obtained from the input texturein a precomputation step. Results from our implementationshow that the algorithm produces high-quality textures,where the time complexity of the synthesis stage islinear in the number of re-synthesized pixels and, therefore,scales well with the size of the input texture.