Real-time cloud detection that uses data parallel and pipeline

Abstract

Remote sensing images play a critical role in many researches such as earth resources investigation, natural disaster forecasting, ground objects identification and climate-related investigations. But the problem of cloud cover limits the application of some remote sensing images. In order to shorten the execution time of the cloud detection and to implement a realtime cloud detection algorithm, this paper proposes a cloud detection algorithm using data parallel and pipeline. By analyzing the computation complexity of Fractal Dimension(FD) and Angular Second Moment(ASM) and analyzing the feasibility of data parallel and pipeline of them, the algorithm using data parallel and pipeline is proposed. The experimental results illustrate that the algorithm using data parallel and pipeline has a good real time.