發明
中華民國
107120602
I 673685
電子影像除霧系統及其方法
元智大學
2019/10/01
霧是指在地面上大氣中懸浮的水汽凝結物,由小水滴或冰晶所組成,是一種常見的天氣現象,霧會影響影像能見度,對交通影響很大,造成大部份車輛均會在大霧時慢駛並使用更多光線照明。馬路上的霧特別危險,很容易釀成意外。 現今有許多除霧演算法來針對此問題做研究,利用物理模型來表現距離和霧生成的相對應關係,針對單一影像的資訊,配合物理模型概念來做除霧能達到一定的除霧效果,影像除霧演算法中,演算流程大致分為四個流程,估算空氣光值、估算透射率、修正透射率、影像還原,也因部分演算法的運算量都偏向於複雜,而在硬體模組實現上較為困難,若要達到即時影像的處理,整個資料量將會非常龐大,如何在高畫質及高幀數的情況中,硬體層面與除霧效果之間取得一個最佳平衡點,本專利使用自行開發的模組來做效率的低延遲影像除霧演算法硬體架構。 The water vapor condensate which suspended on the air, which composed of small water droplets or ice crystals, is a common weather phenomenon called Fog. Fog will affect not only the visibility but also a great impact on traffic. Most of the driver will drive slowly and use fog lights to improve their safety because the fog on the road is very dangerous and easily lead to accidents. Nowadays, there are many defog algorithms to do the relative research which utilize the physical models to represent the corresponding relationship between the distance and the fog. We can easily defog with the physical model for a single image. In the image defogging algorithm, the process is divided into four parts, Airlight Estimating, Transmission Estimating, Refined Transmission, and Scene Recovery. The complexity of partial algorithm is biased towards complicated so that it is difficult to implement the hardware. The amount of data will be very large with the real-time image processing. It is a critical issue to get the best balance point between defog performance and hardware implementation. This patent proposed an efficient image defog algorithms and implement the hardware architectures.
產學合作組
(03)4638800#2286
版權所有 © 國家科學及技術委員會 National Science and Technology Council All Rights Reserved.
建議使用IE 11或以上版本瀏覽器,最佳瀏覽解析度為1024x768以上|政府網站資料開放宣告
主辦單位:國家科學及技術委員會 執行單位:台灣經濟研究院 網站維護:台灣經濟研究院