視覺式交通監控之能見度改善

Abstract

一個可靠的系統，必須要能容忍在不同的環境中(如：起霧、下雨、下雪等)也能確實的運作。本研究最主要的目的，不僅僅是想了解光與空氣是如何的作用。更進一步的，希望藉由公式的推導，回復光在受未干擾時的狀態，也就是景物表面最初所散發的光。 恢復景物原本的影像，無可避免的可以分為兩大問題。分別為估計airlight 與估計透射率 。本研究藉由兩種不同的方法分別得到airlight的值與透射率 值，再藉由影像退化模式的公式推估出景物在未受干擾前最原始的亮度。 實驗的場景我們選擇了在台灣常見的天氣型態：起霧和下雨。並且針對各種的天氣型態拍攝了交通影像與非交通影像。再針對交通影像拍攝了一般道路與高速公路兩種類型。本研究所提出的技術對於能見度之改善皆有良好的效果。The bad weather condition including fog, haze, rain and show make the acquired images contrast being low, which make computer vision application fail to detect or recognize objects. A technique for visibility improvement the bad weather images is presented. In this proposed approach, the polarized images are used to obtain the high contrast images. The input images are assumed composed by the direct light and airlight . The airlight is affected by the bad weather conditions, which makes the acquired image unclear. The key idea of the proposed method is use ICA to estimate and remove the airlight from the input images. The experimental results have demonstrated that the proposed method can effectively remove the bad weather condition and enhance the contrast of the input images