The Research And Realization In Handwritten Fuzzy Identify Method

In the current economic activities, handwritten signatures often replace stamps. Handwritten signatures can be images to identify the authenticity of signatures. The style of a skilled handwritten signature is stability, but font, stroke, size and tilt of a certain arbitrariness, and it should not be identified by printed Chinese character identification methods. The topics used BP neural network memory signature sample characteristics and tectonic vague identifier, the use of VC++ platform achieved in the processing and extraction of features and identifier.Identification system is based on the handwritten characteristics, which are detailed in window standardize and thinning of stroke and the algorithm has its own characteristics.Images used to identify characteristics of density in feature extraction algorithms, taken down victoria technology to reduce the size and improve the speed of BP network.BP algorithms are used to construct characteristics library and vague identifier, improving BP algorithm is the means to enhance the speed of network.This thesis introduces BP algorithm in the application of handwritten signatures vague identification,every details are described in this paper,emphasized on image processing,features extraction, BP fuzzy identification. This system also provide conferences to the fields of handwritten numbers identification and vehicle registration.The thesis is structured as follows :Chapter 1 : introduce the development and study situation of handwritten signature technologies,and put forward the main elements of the research.Chapter 2 : the development condition,the basic principles and its specific application are introduced in this chapter,especially focus on BP algorithms.Chapter 3 : it gives whole design of handwritten signature fuzzy identification system, and detailed in each functional modules.Chapter 4 : research image processing technologies, and providing specifics of various approaches.Particularly skewed calibration and thinning algorithm is dealt with in detail.Chapter 5 : analyzed various features selected algorithms,and compared their advantages and disadvantages, presented the algorithms of this system and describe the methods to achieve this elaborate.Chapter 6 : the principle based on BP neural network is introduced,and this chapter also described sample study and fuzzy identification module, finally it gave a specific realization algorithms.Chapter 7 : examples are described and summed up the research results,finally it pointed out that the inadequate and proposed improvements.