隨著虛擬實境技術發展，360度環景影像已成為沉浸式體驗的重要媒介。然而，現有4K解析度環景影像在虛擬實境頭戴式顯示器中呈現時，因視角分配像素有限而導致畫面模糊或變形，影響使用者體驗。本研究開發整合濾波機制的360 SRGAN損失函數模型，針對等距矩形投影影像解析度不足問題提出解決方案。採用優化的5×5卷積核心處理360度影像特性，並設計360 SRGAN結合銳化濾波器與中值濾波器的損失函數，在提升影像細節的同時有效抑制雜訊。更透過Meta Quest 2進行使用者感知實驗，主觀評估驗證顯示，使用者在對比度和銳利度感知上有明顯改善(p < 0.05)，且不受專業背景影響。本研究為VR環景影像超解析提供技術解決方案，具虛擬實境360環景應用價值。

With the advancement of virtual reality technology, 360-degree equirectangular images have become a crucial medium for immersive experiences. However, when existing 4K resolution equirectangular images are presented through virtual reality head-mounted displays, the limited pixel allocation across the viewing angle results in blurred or distorted visuals, thereby compromising user experience. This study develops a 360 SRGAN loss function model integrated with filtering mechanisms to address the insufficient resolution of equirectangular projection images. The research employs an optimized 5×5 convolutional kernel to process the characteristics of 360-degree images and designs a 360 SRGAN that combines loss functions of sharpening filters and median filters, effectively enhancing image details while suppressing noise. Furthermore, user perception experiments conducted through Meta Quest 2 with subjective evaluation verification demonstrate significant improvements in user perception of contrast and sharpness (p < 0.05), with results unaffected by professional background. This research provides a technical solution for super-resolution of VR equirectangular images and holds substantial application value for virtual reality 360-degree equirectangular environments.