Research On Visible Light Iris Location Based On Image Quality Evaluation

With the rapid development of economy and technology,more and more attention has been paid to the personal information security,which makes the biometric technology develop rapidly and mature gradually.The realization of iris identification under visible light in intelligent mobile terminals is one of the hot research topics.Therefore,this paper studies iris image quality evaluation and iris location method based on the mobile iris recognition system which is suitable for visible light condition.Iris image quality evaluation and iris location are important steps in the iris recognition system.The visible light image obtained by the intelligent mobile terminal is easily interfered and cannot be accurately located.Even if the location is successful,the texture information of the obtained image may not meet the recognition requirements due to lighting problems.In order to solve the above problems,a cascaded frame combining image quality evaluation and iris region localization is proposed in this paper,which is based on the video captured by intelligent mobile terminal and is oriented to multi-frame sequence of iris images.To ensure the accuracy of iris location and the reliability of iris texture information,the proposed framework not only realizes the iris image quality evaluation,but also completes the iris region location.Firstly,the iris image quality is evaluated by Tenegrad clear evaluation function,and the fuzzy image caused by motion or defocus is excluded.Secondly,a HOG +SVM classification model is established to achieve the initial location of the iris,and the iris area is initially located in the human eye image to exclude non-iris areas such as human face and eyebrows.And then the wavelet coefficients of iris region are extracted by wavelet transform,and the feature vector is input into BP neural network to evaluate iris texture quality.The iris image with unclear texture is eliminated after this network.Finally,the inner and outer edges of iris are precisely located by fitting circle method and hybrid geodesic evolution method.The algorithm proposed in this paper is tested with the UBIRIS.v2 iris gallery,which is collected under visible light conditions.The gallery contains various interference images that may be encountered in the practical application of the mobile iris recognition system,which can be used to test the applicability of this method.Nine frame images of 100 people in the gallery are selected to construct 900 iris image sequence for testing which aims to simulate the video sequence images collected by an intelligent mobile terminal.The experimental results show that the iris location accuracy of this method is 94%,and the operating efficiency is improved by 55%.In addition,nine frames images of a video shot by the iphone8 rear camera is used to test the effectiveness of this method.The results show that the method proposed this paper can exclude the disturbed iris image and select high-quality iris images to complete accurate location.