In recent years, dynamic and invisible screen-camera short-range communication for IoT devices have become a popular research field due to its superior user experience and no-extra hardware required. However, most existing screen-camera communication systems suffer from poor data security, high computational overhead, and limited capture angle, which make them infeasible in practice. In this work, we propose MobiScan, a dynamic and invisible screen-to-camera communication system that is able to ensure data security, real-time communication, and flexible capture angle. MobiScan is composed of two parts: a novel fast frame correction scheme and a multilevel data pattern scheme to address the above problems. The fast frame correction scheme proposes a novel frame correct approach that utilizes the sensor data on the smartphone to correct the captured frame by reproducing the relationship between positions of the screen and the smartphone. Through this method, the scheme can save frame correction time without calculating frame content to identify the degree of deformation. The multilevel data pattern scheme protects the data privacy by sending the classified information to targeted users, which can reduce the time overhead and extend the application scenario by redesigning the data structure. The experimental results show that for flexible capture angle problem, the maximum capture angle is 180° in the rotate view and 60° in the side view, bottom view, and overlook view. For real-time communication problem, the time overhead is reduced by 90% in the procedure of the frame correction. In the data decoding process, the time overhead is reduced by 10%. MobiScan enables a flexible capture angle on the PC platform for secure and real-time communication.

中文翻译：

---

MobiScan：用于物联网应用的增强型隐形屏幕摄像头通信系统

近年来，物联网设备的动态和隐形屏幕摄像头短距离通信因其卓越的用户体验和无需额外的硬件而成为热门研究领域。然而，大多数现有的屏幕摄像头通信系统都存在数据安全性差、计算开销大、捕获角度有限等问题，这使得它们在实践中不可行。在这项工作中，我们提出了 MobiScan，这是一种动态且不可见的屏幕到相机通信系统，能够确保数据安全、实时通信和灵活的捕获角度。MobiScan 由两部分组成：一个新颖的快速帧校正方案和一个多级数据模式方案来解决上述问题。快速帧校正方案提出了一种新颖的帧校正方法，该方法利用智能手机上的传感器数据通过再现屏幕和智能手机的位置之间的关系来校正捕获的帧。通过这种方法，该方案可以节省帧校正时间，而无需计算帧内容来识别变形程度。多级数据模式方案通过将分类信息发送给目标用户来保护数据隐私，通过重新设计数据结构可以减少时间开销并扩展应用场景。实验结果表明，对于灵活捕捉角度问题，最大捕捉角度在旋转视图中为180° ，最大捕捉角度为60° 。在侧视图、仰视图和俯视视图中。对于实时通信问题，帧校正过程中的时间开销减少了90%。在数据解码过程中，时间开销减少了10%。MobiScan 可在 PC 平台上实现灵活的捕获角度，以实现安全和实时的通信。