Agriculture Parcel Boundary Detection
from Remotely Sensed Images
Ganesh Khadanga and Kamal Jain
Abstract The object-based image analysis (OBIA) is extensively used nowadays
for classification of high-resolution satellite images (HRSI). In OBIA, the anal-
ysis is based on a group of pixels known as objects. It differs from the traditional
pixels-based methodology, where individual pixels are analyzed. In OBIA, the image
analysis consists of image segmentation, object attribution, and classification. The
segmentation process thus identifies a group of pixels and are known as objects.
These objects are taken for further analysis. Thus segmentation is an important step
in OBIA. In order to find out the boundary of agriculture parcels, a two-step pro-
cess is followed. First, the segmentation of the images is done using the statistical
region merging (SRM) technique. Then the boundary information and center of the
segmentation are found out using MATLAB. The best fit segment was found out
using trial and errors. The extracted boundary information is very encouraging and it
matches the parcel boundaries recorded in revenue registers. The completeness and
precision analysis of the plots are also quite satisfactory.
Keywords MATLAB · Segmentation · Cadastral parcel · OBIA · HRSI
1 Introduction
The object-based image analysis (OBIA) is extensively used nowadays for classifica-
tion of high-resolution satellite images (HRSI). In the traditional pixel-based analysis
each pixel is taken into account and the classifications are done based on the spectral
characteristics of the pixel. But in the object-based analysis, a group of objects is
identified based on certain homogeneity condition [1]. This group of homogeneous
pixels is known as an object. These objects were taken up for classification of the
features in the image. Thus the OBIA analysis is based on this group of pixels. Dur-
ing the classification stage the contextual relationship like neighborhood, spatial, and
G. Khadanga (B ) · K. Jain
Civil Engineering Department, Geomatics Group, Indian Institute of Technology Roorkee,
Roorkee, India
e-mail: ganesh@nic.in
© Springer Nature Singapore Pte Ltd. 2020
B. B. Chaudhuri et al. (eds.), Proceedings of 3rd International Conference on Computer
Vision and Image Processing, Advances in Intelligent Systems and Computing 1022,
https://doi.org/10.1007/978-981-32-9088-4_26
307