Research On Finger Vein Recognition System Based On Micro Lens Array Ultrathin Device

Biometric identification technology uses different physiological characteristics of human for personal identification,such as fingerprints,face,palm prints,etc.As each person's biological characteristic is unique with high security,it is widely employed in checking work attendance,police criminal investigation,security and other fields.Among many biometric recognition methods,finger vein recognition technology has become research hotspot because of its internal characteristic and living body recognition.However,the existing finger vein recognition technology limited to optical imaging is inconvenient due to its large volume.With the development of mobile platforms,such as cellphone,the problem that needs to be solved is that how to apply the identification system to the mobile field.Meanwhile,the current research focus is how to greatly improve its recognition.On the basis of the artificial compound eye formed by micro lens array,we constitute the complete finger vein image by the collection of series of image sequences and its fragment assembly.After that,we conduct the extraction and recognition of the image features.In this paper,firstly we introduce the principle of acquiring images and the hardware structure of the system.Then,the introduction is the stitching algorithm of finger vein sequence which is one of the systematic nucleuses.In this step,we first search the best overlap area on account of template matching to position the overlapping portion of finger vein between the two adjacent images.Then we can obtain a complete vein image through the image fusion according to the overlapping part.There will be the ROI extraction process as well as the preprocessing before the feature extraction of image.During the former,one of the key points is image rotation correction.It is conducted based on the least square fitting algorithm by the finger edge lines.At last the correction receives the good results.The other center of this paper is the feature extraction and recognition of finger vein.In this part some commonly feature extraction methods are introduced,such as the way based on multi-threshold LBP coding,of which the principle and the realization process is the key points.The main contributions of this paper are as follows:1).We have collected the meridian image by micro lens array and have conducted its fragment assembly.2).We have solved the problems,such as the ROI extraction of finger displacement.3).We have proposed the feature extraction method for multi-threshold LBP finger vein.Finally,compare to the results of other recognition algorithms in the self-built database,the experiment shows that the algorithm in this paper posses the higher recognition efficiency.