J Jacinth Jennifer
PG student, Department of Civil Engineering, Anna University Regional Campus Tirunelveli, India
S Vanmathy
PG student, Department of Civil Engineering, Anna University Regional Campus Tirunelveli, India
C Maria Jobi Sahana
PG student, Department of Civil Engineering, Anna University Regional Campus Tirunelveli, India
G Devi
Assistant Professor, Department of Civil Engineering Anna University Regional Campus Tirunelveli, India
Download PDFRemote sensing technology is paving its own way with remarkable advantages in various fields. Feature extraction from satellite imagery is one of the challenging tasks in image processing. As far as low resolution images were concerned, per pixel analysis and sub-pixel analysis were gaining its importance. Now with the advancements in technology, high resolution imageries are acquired easily and made used in various applications. When speaking about high resolution imageries, it is made known that an object or feature in the imagery is made up of several pixels which is contrary to the low resolution imageries. Thus, an alternative technique for feature retrieval is necessary to extract features from high resolution imagery. Spatial relationships between the pixels were taken into consideration along with its spectral characteristics. Texture is one of the spatial parameter which is of much importance. The texture of an image gives us the information about the spatial arrangement of colours or intensities and it is a function of the texture surface, its albedo, the illumination and the camera and its viewing position. There are various parameters to characterize the texture of an image. Haralick’s texture parameters were found to be of much importance compared to the other texture parameters. Thus Haralick’s texture parameters were considered in the study. There are about thirteen Haralick’s texture parameters. In urban feature extraction, it is not necessary that all these thirteen parameters have to be imposed because certain parameters have no influence in extracting the urban features. So based on this aspect, statistical analysis was made so as to examine and quantify the influence of each Haralick’s texture parameter. Out of thirteen, six were found to be of considerable importance. Using these six textural characteristics, classification was carried out and 88% accuracy was obtained in urban feature extraction.
Keywords: texture; satellite imagery; haralick’s texture parameters; statistical analysis; urban feature extraction; classification
Disclaimer: All papers published in IJRST will be indexed on Google Search Engine as per their policy.
