Detection of ponding in Satellite Images

摘要

We describe a system for the identification of Logging in a field from Image Data. In one approach, analyzes an image of an agricultural Field using a classifier that has been trained on the basis of Spectral Bands of pixels labeled images to identify within a probability for each pixel in the image where the Pixel corresponds to water.We performed a simulation to determine regions in the flow Field where it is likely that hold Water After Rain based on precipitation Data, Elevation Data and Property Data of field soil. It generates a graph of vertices representing the pixels and edges represent the connections between neighboring pixels.The probability of each pixel within the graph represents ponding is adjusted based on the likelihood that the Pixel represents Water, the possibility that the water accumulates in the area represented by the pixel, the probability that neighboring pixels are accumulations of Water And a Mask of arable land that identifies the pixels that correspond to crop Land.Then it determines a class for each pixel that maximizes the Joint probability in the figure.