关键词: 图像预评估, 实时虹膜识别系统, 联合评估网络, 坏样本

Abstract: There are different types of bad images when an iris identification application system captured iris images.It is well known that these bad images usually result in localization or identification errors and increase the failure rate to enrollment if they are imported into the identification processing.Because previous image quality evaluation methods estimate it whether bad or else by the resolution and the definition of the iris part after having calculated the iris location of an input image,they just can deal with few types among them.This paper proposes an idea of real-time pre-estimation using the associated estimation network,based on the detailed and objective analysis of reasons resulting in the bad-image,in which it pre-estimates images temporarily saved in memory before the localization or rough localization process and decides the system whether captures an iris sample again or turns into the next step by the output of the network.The experimental result is shown that the method can detects most types of the bad-image with comparatively low false rate and fairly large throughput.It should satisfy the pre-estimation requirement of a real-time iris identification system.

Key words: image pre-estimation, real-time iris identification system, associated pre-estimation network, bad-image