Iris is a unique biometric characteristic of the human body,making it suitable for individual i-dentification.However,in the field of justice,the results of iris recognition algorithms have yet to be ac-cepted as evidence.Recently,the forensic science community has begun to address the issue of iris iden-tification,proposing the examination and identification through the visible texture features of iris.To solve the feature matching problem in iris forensic identification,an interpretable expression method for block-shaped iris features is proposed.The RAA-UNet segmentation model is employed to extract iris block-shaped features,which are then described and compared in terms of shape and position.The intra-class differences of block-shaped iris feature matching are quantified by establishing an intra-class block-shaped iris feature dataset.Experimental results demonstrate that the descriptive information of intra-class block-shaped iris features can remain relatively stable,supporting the feasibility of the proposed method for identifying block-shaped iris feature matching.Furthermore,the range of descriptive parameters dur-ing feature matching is obtained,which provides methods and statistical data support for future research on irises from different individuals with similar features.
万方数据
