Fine-grained identification of camera devices based on inherent features
2022
Ruimin Wang | Ruixiang Li | Weiyu Dong | Zhiyong Zhang | Liehui Jiang
Camera devices are being deployed everywhere. Cities, enterprises, and more and more smart homes are using camera devices. Fine-grained identification of devices brings an in-depth understanding of the characteristics of these devices. Identifying the device type helps secure the device safe. But, existing device identification methods have difficulty in distinguishing fine-grained types of devices. To address this challenge, we propose a fine-grained identification method based on the camera deviceso inherent features. First, feature selection is based on the coverage and differences of the inherent features type. Second, the features are classified according to their representation. A design feature similarity calculation strategy (FSCS) for each type of feature is established. Then the feature weights are determined based on feature entropy. Finally, we present a device similarity model based on the FSCS and feature weights. And we use this model to identify the fine-grained type of a target device. We have evaluated our method on Dahua and Hikvision camera devices. The experimental results show that we can identify the deviceos fine-grained type when some inherent feature values are missing. Even when the inherent feature pmissing rateq is 50%, the average accuracy still exceeds 80%.
显示更多 [+] 显示较少 [-]