A combined multi-scale/irregular algorithm for the vectorization of noisy digital contours.
Antoine VacavantTristan RoussillonBertrand KerautretJacques-Olivier LachaudPublished in: Comput. Vis. Image Underst. (2013)
Keyphrases
- multiscale
- learning algorithm
- dynamic programming
- computational cost
- noisy data
- matching algorithm
- cost function
- search space
- computational complexity
- preprocessing
- neural network
- worst case
- objective function
- detection algorithm
- k means
- np hard
- optimal solution
- high accuracy
- input data
- segmentation algorithm
- digital straight line
- image processing
- clustering method
- theoretical analysis
- natural images
- expectation maximization
- simulated annealing
- probabilistic model
- experimental evaluation