We are researching a new display device that allows the public to experience augmented reality without wearable devices. In this paper, we implement an image processing system that improves the dynamically changing image quality deterioration, which is a problem unique to optical systems that project images on the real space, and specify the necessary system requirements from the results of verification of the effect. We quantify the characteristics of blur for each of the three primary colors defined as digital image data, and implement image processing that applies filter correction to the input image data during aerial imaging. Since the effect on image quality depends on what kind of imaging optical path is formed, we designed the device structure assuming that it will be applied to signage products, and built an experimental environment that can analyze the aerial image that the user actually sees. The image subjected to the correction process follows an optical path that forms an image in the air, and the user visually recognizes the image with emphasized edges. We compared the results of the actual aerial image and the simulation results, and clarified the system requirements for realizing an image processing system with higher correction accuracy.

我们正在研究一种新的显示设备，让公众无需可穿戴设备即可体验增强现实。在本文中，我们实现了一种图像处理系统，该系统可以改善动态变化的图像质量劣化，这是将图像投影到真实空间的光学系统所特有的问题，并从效果验证的结果中指定了必要的系统要求。我们量化定义为数字图像数据的三基色中每一种的模糊特征，并实现在航空成像期间对输入图像数据应用滤波器校正的图像处理。由于对图像质量的影响取决于形成什么样的成像光路，因此我们设计了设备结构，假设它将应用于标牌产品，并构建了一个可以分析用户实际看到的航拍图像的实验环境。经过校正处理的图像沿着在空气中形成图像的光路，用户视觉上识别边缘被强调的图像。我们将实际航拍图像的结果与仿真结果进行了比较，明确了实现更高校正精度的图像处理系统的系统要求。