Research And Development For A Fingerprint Preprocess System

With the rapid development of modern society and the coming of information age, biometric technology is becoming more and more popular in the field of personal identification and the research of automatic fingerprint identification is the most remarkable. Nowadays, automatic fingerprint identification technology has been applied to these domains such as police, CIQ and bank etc. As the rapid developing of computer network, automatic fingerprint identification technology additionally provides a new method which can tackle the safety of network and database.The fingerprint recognition technology first step is the fingerprint gathering work, how could gather the clearer complete fingerprint image or how can carry on not the clear fingerprint image return to original state processing to become very much the key point. The fingerprint feature extraction algorithm's performance must rely on to a great extent fingerprint image own quality. Specially regarding the quality good fingerprint image, may monitor its grain line easily from the refinement image, the characteristic point may the very quick pinpointing. But, in actual application process, because the gathering condition and gathers equipment's factor, gathers some fingerprint image has many are the quality is quite bad, easy to cause the following several questions: (1) Possibly produces the massive false characteristic points; (2) Loses the massive true characteristic points; (3) The characteristic point localization (position and direction) possibly have the big deviation. Uses such feature extraction result is carries on the fingerprint match correctly very difficultly. Therefore, must study the fingerprint the enhancement technology, guaranteed that the feature extraction algorithm has enough processing regarding the image quality, regarding the automatic fingerprint processing technology is very essential. From a other angle, is to the quality relatively quite bad fingerprint, the fingerprint expert may also use such as the partial grain line direction, grain line continuity and grain line visual cues and so on tendency distinguishes the fingerprint many characteristics, can thus carry on the fingerprint correctly the match. Therefore, theoretically, developed one kind to be able to use these visual cues to strengthen the fingerprint the picture quality technology is also fully very feasible.Image place Neo-Confucianism is a comprehensive edge discipline, converged in image place Neo-Confucianism the gamp under optics, electronics, mathematics, the camera technique, the computer technology and so on numerous aspect disciplines. Looking from the research scope, it with the computer graphics, the pattern recognition, the computer vision and so on already has contacting also to have the difference.The image smoothing is in the imagery processing key technologies. Eliminates the image noise the work to be called as is the image smoothing or the filter, is one kind of very practical digital image processing technology. Because actual factor and so on image formation condition and environment influences, usually absorb the image was in varying degrees already has polluted the image, in the image noise and the signal interweaves mutually in the same place, particularly rode the noise, if processed smoothly improper, will cause image's detail like boundary outline, the line and so on will become slurred. Therefore, when carries on the image smoothing, must choose the appropriate method. Because the noise source are many, the noise type is complex, therefore the smooth method is also many and varied. Was opposite in the majority images, the people will not know or impossible to describe the signal and the noise model precisely with the simple stochastic process, since therefore, period of time the people have used do not need to know any have about the image apriori knowledge fixed window spatial domain smooth algorithm, this algorithm is no doubt very simple, but they will cause the image during noise reduction to produce fuzzily. Therefore the people also proposed many improvement's image smoothing algorithms, obtain during noise reduction maintain as far as possible the image the detail information, the value filter are also one kind. The present paper uses the above two algorithms, namely and the value filtering regarding carries on based on the Gauss template's weighting neighborhood method of average by the spiced salt noise and the stochastic noise pollution's image processing, and carries on two methods the contrast, makes every effort to discover the respective good and bad points.We to two partial feasibilities which proposed in the question definition phase carry on the analysis. First is the artificial programming makes the stochastic noise and the spiced salt noise, these two kind of noises are the internal noises, is by reasons and so on light and electricity basic property, mechanical movement, primary device noise, system internal circuit noise causes, moreover the statistical nature along with the time variation, is not the stationary noise, the man-power simulates does not have the problem. What method is the next question selects to carry on the image smoothing and the filter. This question belongs to an image intensification part, already had many classics the algorithms, like the partial smoothing procedures, ultra limit the picture element smoothing procedures, a gradation closest X neighbour method of average, the gradient from the bottom weighting smoothing procedures, the biggest uniformity smooth, to have the choice to guarantee the edge smoothing procedures, the spatial low pass filtering, many image methods of average, value filter and so on, therefore which one method effect solves this problems is only uses to be better, perhaps which method is suitable for the stochastic noise, which kind gathers the solution spiced salt noise.The paper is a part of automatic fingerprint identification system; it discusses and studies a number of items about the system. We introduce several Image smoothing methods based on random noise and pepper-salt noise which is named Gauss mask image smoothing and median filtering. In fact, the quality of the image which will be processed is pretty good, so we add some noise to it artificially with the purpose of simulating the noise caused by nature and then do the smoothing work to the image of poor quality in order to make an obvious effect. The algorithm can enhance the quality of fingerprint image, improve the clarification of ridge and furrow structure of fingerprint image of poor quality and then improve the accuracy of feather extraction.The steps of writing this thesis basically depends on the requirement of every period of software project. It totally contains five episodes and what content in each of them is: The first chapter is the introduction, what main narration is this topic overall outline and the significance and the goal; The second chapter for image pretreatment technology, what main elaboration is the image pretreatment technology principle and the denoising algorithm; The third chapter: What theoretical analysis and plan main introduction programs tool VC++; The fourth chapter is the system planning and realizes the process in detail, is overall system's demand analysis and the module design and the algorithm realizes the process; The fifth chapter is the conclusion, summarized each algorithm comparison as the paper ending.