Research On The Feedback Intelligent Cognitive Model And Its Operation Mechanism For Offline Handwritten Chinese Character Image With Uncertain Result Performance Index Constraint

Offline handwritten Chinese character image recognition is a hot area of research in the field of artificial intelligence,with a wide range of theoretical research value and applicable prospect.Due to the characteristics of writing style of offline handwritten Chinese character images,aimed at the unsatisfied performance from the traditional single-layer and unidirectional open-loop cognitive pattern,the human cognitive pattern with the characteristics of repeated comparison and inference from global to local is imitated in this dissertation,to explore an offline handwritten Chinese character image intelligent cognitive model with feedback adjustment mechanism based on multi-cognitive method fusion and uncertain result performance index constraint,to improve the cognitive precision for offline handwritten Chinese character image.The main contributions of this dissertation are list as follows:(1)Research on the intelligent cognitive decision information system model of offline handwritten Chinese character image.Specific to the no unified description problem of offline handwritten Chinese character image decision information system,based on rough set theory,the real decision attribute information of offline handwritten Chinese character image is introduced in this dissertation,to establish an offline handwritten Chinese character image cognitive intelligent decision information system model with multi-level feature representation from the perspective of information theory.(2)Research on the intelligent cognitive system architecture of offline handwritten Chinese character image.Based on the structural and statistical feature extraction methods,random vector functional-linker(RVFL)network ensembles,and fuzzy integral,the multi-level feature space data structure,the classified cognitive criterion with ensemble structure,and the fused cognition method of offline handwritten Chinese character image are acquired,to overcome the drawbacks of fixed one-sided feature space and classified cognitive criterion from traditional single method cognitive system.(3)Research on the definition and calculation model of uncertain cognitive result evaluation system.Aimed at the problem that current posterior probability assessment method can't evaluate the cognitive result credibility online,based on generalized error theory and generalized entropy theory,the human cognitive process from rough comparison to detailed comparison is imitated,to define a generalized cognitive result similarity entropy measure index,to provide a quantitative basis for feedback cognitive intelligent operation mechanism.(4)Research on the feedback cognitive intelligent operation mechanism with uncertain cognitive result measure index constraint.Aimed at the problem that significant difference exists between the traditional unidirectional open-loop cognitive pattern and the human cognitive pattern with the characteristics of repeated comparison and inference from global to local,an offline handwritten Chinese character image cognitive intelligent operation mechanism with feedback pattern and uncertain cognitive result performance measure index constraint is constructed,and the adjustment mechanism of multi-level feature space inter-layer regulation and intra-layer optimization is established based on the generalized cognitive result similarity.In order to verify the superiority of the proposed intelligent cognitive model,GB23122-80 standard simplified handwritten Chinese character library is used in this dissertation by MATLAB simulation.The experimental results show that the average recognition rate reached 93.78%,which is better than the traditional open-loop cognitive models and closed-loop cognitive model based on fixed methods,and establishs the foundation of “machine recognition” for the offline handwritten Chinese characters.