High-level Markovian model-based low-quality Chinese character initial framework connecting algorithm

Abstract

The invention discloses a high-level Markovian random field model (High Level MRF)-based low-quality Chinese character initial framework connecting algorithm, relating to a computer image processing technology. The high-level Markovian model-based low-quality Chinese character initial framework connecting algorithm combines optimization and random field theories, a high-level MRF model is connected with an initial framework, prior information (such as angle difference, translation difference and end point distance) of Chinese character frameworks and structures can be added as much as possible on the basis of the high-level MRF to be used as the restriction, and therefore, the marking problem is optimally solved. The high-level Markovian model-based low-quality Chinese character initial framework connecting algorithm provided by the invention is also capable of obtaining a good framework even if in the case that the framework is seriously broken so that the accuracy of a low-quality Chinese character framework is improved, and the algorithm can be applied to the technical fields such as Chinese character information processing, office automation, machine translation and artificial intelligence.

Description

technical field [0001] The invention relates to computer image processing technology, in particular to an initial skeleton connection method for low-quality Chinese characters. Background technique Because low-quality Chinese characters are affected by various degrading factors, its skeleton extraction is more difficult than that of ideal Chinese characters. Even if the initial skeleton can be extracted, it is unavoidable that the skeleton is interrupted, the skeleton of individual strokes is missing, and it does not conform to human vision. If Connect these broken skeletons to make them conform to human vision, then the problem of low-quality skeleton extraction is easily solved. There are two existing methods for connecting skeletons, one is the shortest Hamiltonian path method, and the other is the singularity analysis method. The Hamiltonian path was proposed by the astronomer Hamilton, which means that for a given network, after determining the starting point and the...

AI Technical Summary

Problems solved by technology

In short, so far, the skeleton connection of low-quality Chinese characters is still a challenging problem, especially for low-quality Chinese characters that are sparse, broken, and noisy. Most existing skeleton extraction algorithms cannot extract skeletons that conform to human vision. i.e. not getting a skeleton that satisfies the criteria for a "good" skeleton

Images