Object Recognition in Earth Surface Satellite Images Using Digital Image Processing and Machine Learning Techniques with Big Data Technologies

Abstract

Detection of an object from a satellite image is a difficult process because the presence of objects in a satellite image is unpredictable. Different approaches have been available to detect vehicles, buildings, trees however all these objects were detected individually through machine learning and some other methods. Similarly accuracy in object detection is another major issue. In our proposed work, To analyze the object accurately, Polygon approach is implemented which includes both shape and color as input and processes it with datasets to attain maximum accurate result. Here image parameters have been extracted accurately through feature detection. After segmentation of a particular object from image CNN classification is implemented. Through this, in our proposal we are going to detect roads, trees, buildings, waterway and few other objects accurately with this single approach.