Research On Character Recognition Feature Extraction Based On Genetic Algorithm

With the continuous development of information society,human beings use characters to exchange information.The society needs a wide range of handwritten character recognition and application.Converting paper data into electronic information plays an important role in an efficient and convenient information society.For offline handwritten characters,the style of handwritten Chinese characters is arbitrary,unrestricted and the structure of strokes is complex.Therefore,handwritten Chinese characters are a challenging task for recognition pattern.The difficulty of character recognition lies in how to extract the most distinguishing features.Traditional statistical feature extraction based on statistics can achieve better results,but the premise is that the training set must be large enough and the training times enough to get the desired effect.However,the structure of strokes remains unchanged in handwritten characters.Although the styles are quite different,the structure of strokes is unique.Therefore,stroke structure features are more effective and accurate for character recognition.In this paper,we want to solve the problem that the result of feature extraction is not stable,so we improve on the basis of stroke structure features.The curve result after the initial character is imaged is transformed into curve straightening,and DDA algorithm is used to solve the influence of multiple inflexion points on feature extraction.Then,the minimum deviation method is used to calculate the error values of input characters and sample sets in the spatial coordinate system,and the deviation values of angles and lines are used for character matching.The minimum deviation value has the highest matching degree.Finally,the genetic algorithm is introduced to iterate the feature set of angle and line segment,and finally get the optimal feature in the process of cross evolution of generations,so as to improve the recognition accuracy.In order to ensure the effect of the experiment,this paper selects the hcl2000 Chinese handwritten character library,and takes the character library as the input item to carry out the experiment by average grouping,puts the feature set into the position matrix,and compares it with the 3755 samples in the database one by one.Finally,it is found that the improved stroke structure feature extraction algorithm based on genetic algorithm can improve the recognition accuracy,better complete,and the effect is better.