Accuracy Assessment of Cloud Reconstruction Approaches Using Segmentation

Cloud is the major obstacle to analyze data in the satellite images. The various approaches are used to remove the cloud from the satellite image for further processing. The approaches are in-painting and multi-temporal. But, the algorithm working for these approaches cannot produce the accurate results. So, that the accuracy assessment helps to motivate the increased accuracy result. The main aim of this paper is to analyze the accuracy of in-paint and multi-temporal approach and produce the pros and cons of those approaches. Accuracy assessment helps to obtain degree of truthfulness of the results. There are 'n' numbers of metrics are available to find the accuracy of the result such as analyzing variance, spatial error, probabilistic error etc. In this paper, two approaches are implemented and the results are applied to the segmentation algorithm. Then, the segmentation results are analyzed by using the error matrix. The error matrix have constructed based on the difference between the clusters of the image result. For segmentation the K-Means algorithm is used and for simplicity only two clusters are segmented. Segmentation result will clearly show that the accuracy of the in-paint and multi-temporal approaches. From the result it is evident that the multi-temporal approach produces a better result when compared to the in-painting. Especially, in that multi-temporal the Averaging method produces better accuracy result.