Satellite Image Segmentation Using Thresholding Technique

Image segmentation is one of the basic techniques of image processing and computer vision. It is a key step for image analysis, comprehension and description. Among all the segmentation techniques, thresholding segmentation method is the most popular algorithm and is widely used in the image segmentation field. The basic idea of automatic thresholding is to automatically select an optimal or several optimal grey-level threshold values for separating objects of interest in an image from the background based on their grey-level distribution. Image segmentation techniques were widely used in image analysis for various areas such as biomedical imaging, intelligent transportation systems and satellite imaging. A major challenge for image segmentation is to segment the complex images with noise, intensity inhomogeneity, texture or multiphase structure. However the main issue in remote sensing is image classification that required determining an appropriate threshold between species in producing accurate segmentation image. Image segmentation on satellite imagery is a complex process and requires consideration of accurate classification system. A pixel in the satellite image may possibly cover more than one object on the ground. A threshold has to be set to classify an overlap of two or more associated spectral properties. Therefore the aim of this study is to determine the optimal threshold value for object classes to ensure the misclassification of image pixels kept as low as possible by analyzing the classification of satellite images at different hierarchical level. Then the optimal threshold value will be proposed on satellite image segmentation for Universiti Teknikal Malaysia, Melaka (UTeM) area. An evaluation on the accuracy of the enhanced threshold value in identifying and classifying the urban objects shall be made. A hierarchical threshold is expected to significant improvement result on an image segmentation final image for UTeM area