Learning Based Visibility Measuring with Images

摘要: Visibility is one of the major items meteorological observation. Its accuracy very important to air, sea and highways transport. A method visibility calculation based on image analysis learning introduced in this paper. First, effectively represented by contrast vision features. Then, a Supported Vector Regression (SVR) system constructed between features target visibility. Consequently, can be measured directly from single inputting with system. The makes use existing video cameras calculate real time. Specific experiments show that has characteristic low cost, fast calculation, convenience. Moreover, our proposed technology used anywhere measure