Pixel based Classification of Poultry Farm using Satellite Images
Roshini M1, V Poompavai2, K Ganesha Raj3
1Roshini M, Department of Electronics and Communication Engineering, PES University- RR Campus, Bengaluru (Karnataka), India.
2Dr. V Poompavai, RRSC- South, NRSC, ISRO Marathahalli, Bengaluru (Karnataka), India.
3Dr. K Ganesha Raj, RRSC- South, NRSC, ISRO Marathahalli, Bengaluru (Karnataka), India.
Manuscript received on 23 November 2019 | Revised Manuscript received on 17 December 2019 | Manuscript Published on 30 December 2019 | PP: 1-6 | Volume-9 Issue-1S5 December 2019 | Retrieval Number: A10011291S52019/19©BEIESP | DOI: 10.35940/ijeat.A1001.1291S519
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Remote sensing has emerged as a compelling tool to survey and monitor natural resources and other features of an area due to the inherent advantages of synoptic view, repetitive nature and capability to study inaccessible areas. Satellite data/aerial photos are interpreted using keys such as colour/tone, texture, pattern, association, size, shape, etc., and computer-based techniques. Presently geospatial technology is used in various sectors like agriculture, forestry, geology, marine, urban and rural planning and so on, with applications in agriculture seeing a rise in India. This paper elaborates on the method employed for identification of poultry farms in India, using images from satellites such as CARTOSAT and RESOURCESAT (LISS4) and also Google Earth Images. Each poultry farm varies in the size and number of poultry sheds which further depend on the number of chickens bred, location of vegetation and water resources nearby, temperature and humidity of location, etc. Thus, based on these factors, training sites in Hessarghatta, Harohalli, Dommasandra near Bengaluru City, Karnataka were identified. The paper elucidates application of vegetation and water masks using the classification of NDVI. Two pixel-based classification techniques – Maximum Likelihood Classifier and K-Nearest Neighbour Classifier using SNAP Application were applied. Statistics were observed for the accuracy of classified output, and it was shown that Maximum Likelihood Classifier provided more accurate results. The method presented in this paper can be fine-tuned and applied for poultry farms anywhere by studying Poultry Farms in different terrains and using various associations to identify them.
Keywords: Remote Sensing, Poultry Farms, Pixel Based Classification.
Scope of the Article: Classification