Design and Analysis of Improved Deep Convolutional Neural Network Based Satellite Remote Sensing Image Segmentation System

Abstract

In recent years, machine learning techniques have shown promising results in image segmentation tasks. Machine learning algorithms can learn complex relationships between input features and output labels, and thereby, improve the accuracy and efficiency of segmentation. There are several machine learning algorithms that can be used for image segmentation, such as supervised, unsupervised, and semi-supervised learning. Supervised learning involves training a model on labeled data, where each pixel in the image is assigned, a label indicating the class it belongs to. Unsupervised learning, on the other hand, does not require labeled data and groups pixels into clusters based on their similarity. Semi-supervised learning is a combination of both supervised and unsupervised learning, where a small portion of labeled data is used to guide the clustering process. Or also label them. In the first case, some implementations do not require training. This research aims at developing improved satellite image segmentation technique for segmentation and labeling of satellite images based on machine learning and deep learning techniques. The results have been compared on the basis of figure of merits such as precision, recall, f1 score and analysis proves that the proposed hybrid technique outperforms traditional machine learning and deep learning techniques. The research may be used for analysis and applications related to surveillance, remote sensing and analysis of important parameters such as deforestation and climate change.