High-speed and high-accuracy algorithm for cloud detection and its application

Abstract

The cloud cover is an important factor which lowers the remote sensing image quality, so real-time automatic cloud detection and effective rejection of high cloud coverage pictures are of prime importance. In this paper we proposed a high performance and high accuracy algorithm of cloud detection which combines two different analytical techniques: the spectrum threshold comparison and the texture analysis. These two approaches discriminate the image from different visions. A structure of the discrimination tree is proposed to improve the accuracy and to accelerate the detecting procedure, which defines the rule how to use these two methods properly. The cloud detection results gained by this algorithm are well satisfied. And the structure of the discrimination tree promotes the operating efficiency on average. We also proposed an advanced approach to calculate the fractal dimension value, which is about five times faster than the original approach. The cloud detection algorithm has been applied to the data processing system of China-Brazil Earth Resources Satellite-2B. The experimental results show that this algorithm can satisfy the demand of error rate: the false alarm rate is lower than 5% and the missed detection rate is lower than 10%.