CAJ | 학술논문

针对在雾霾天气下无人机航拍的视频图像对比度和色彩保真度差等问题，传统的暗原色先验方法虽然有较明显的去雾效果，但算法复杂度高、耗时长、无法满足实时性需求。本文提出一种基于金字塔技术的快速去雾方法，首先利用高斯金字塔将原始有雾视频图像降采样，然后将降采样后的图像通过暗原色先验的方法去雾处理，最后再采用拉普拉斯金字塔将结果图升采样为无雾视频图像。通过实验结果表明，本文方法在保证去雾效果的同时，提升了场景复原的速度。针对航拍视频场景，能较好地满足实时性要求。
침대재무매천기하무인궤항박적시빈도상대비도화색채보진도차등문제，전통적암원색선험방법수연유교명현적거무효과，단산법복잡도고、모시장、무법만족실시성수구。본문제출일충기우금자탑기술적쾌속거무방법，수선이용고사금자탑장원시유무시빈도상강채양，연후장강채양후적도상통과암원색선험적방법거무처리，최후재채용랍보랍사금자탑장결과도승채양위무무시빈도상。통과실험결과표명，본문방법재보증거무효과적동시，제승료장경복원적속도。침대항박시빈장경，능교호지만족실시성요구。
In view of video images shot by drones having poor contrast and color fidelity under foggy and hazy weath-ers,traditional dark channel priority method has complex algorithm and is time-consuming,which can not meet the re-al-time requirement,although it has obvious fog removal effects. A fast defogging method based on Pyramid technolo-gy is proposed in this paper.Firstly, the original foggy image will be down-sampled by gaussian pyramid. Then, the sampled images are degogged with the dark channel priority method. Finally, the processed images are up-sampled to fogless video images with the laplacian pyramid. The experimental results show that the method can increase the scene recovery speed on the basis of good defogging effect. For aerial video scene, the method can meet the real-time re-quirement.