Nonlinear Image Enhancement Based Localization For Nonideal Iris

With the rapid development of information technology, the importance of informationsecurity is more and more prominent, the security of personal information and accuratealso put forward higher request, therefore biometric technology attaches greatimportance to people.Traditional biometric identification technology, such as fingerprintrecognition.Due to its own defects has been unable to meet the high demand for somesecurity industry. As a new recognition,iris recognition, because of its uniqueness,stability, security features has being recognized as one of the most safe and reliablebiometric technology [1], at the same time, because of its non invasive makes it has avery broad market prospect in areas such as public security and social management.Sothe iris recognition technology has great significance to information security in today’ssociety.A typical iris recognition system includes image acquisition, image preprocessing,feature extraction (code), feature matching. Mosttines, image acquisition uses CCDcamera to get iris images directly, the sharpness of the image has a decisive role for irislocalization;Iris image preprocessing is the most important step in iris recognitionprogram, and iris boundary localization is the key to image preprocessing, especially itsaccuracy has important influence to the result of the recognition. generally according tothe characteristics of the iris shape, people model iris boundary as a ring. it includesboth inside and outside boundarys; Iiris feature extraction is extracted usefulinformation and coding from the preprocessed iris image; Feature matching is toclassify iris image matching identification using the similarity measure.This article willmainly carried out on the inside and outside iris boundary localization.In this paper we propose the boundaries localization for non-ideal iris images based onthe nonlinear image enhancement. For non-ideal iris images, frequently-occurred strong texture, low contrast between iris and sclera, and boundary can cause very strongpseudo border after using canny operator, that will lead inaccuracy localization of irisboundaries, particularly the outer one. In order to solve these problems in this paper, thenon-linear transform is made for the iris image, weaken the iris internal texture, averagethe gray at the same time to enhance the iris outer boundary of the iris. Thus weakeningthe iris internal texture and enhancing the boundary of the iris, make it more conducivetolocation. In order to verify the effectiveness of the proposed algorithm, we dosimulation experiment using the Chinese academy of sciences institute of patternrecognition.CASIA-Iris database. From the analysis of the proposed algorithm results,it can effectively improve the non-ideal iris localization accuracy.In addition this paper summarizes we improve in the the wildes iris localizationalgorithm on the basis of the existing iris localization algorithm.Wildes iris recognitionsystem uses in the canny operator to do edge detection and use circle fitting houghtransform methodto do iris localization, but for non-ideal iris because of the influence ofthe internal pseudo iris boundary, and use the most edge points standard hardly locatethe real boundary of the iris, the actual location will be a little error.So in this articlethreshold value method is used to Get rid of the minor pseudo border, and improve thehough transform, make the most points fell around the border. The experimental resultsdepict that our algorithm improves the non-ideal iris localization accuracy...