Abstract
Since extraction of useful information from remote sensing data is important, scientists manage to propose efficient algorithms for automatic extraction of constructive information from the satellite imageries. To date, image classification has benefitted from advancements in improved computational power and algorithm development. Therefore, Satellite image classification using GeoEye-1, High Resolution Satellite Image (HRSI) of 2016, Support Vector Machine Classifier (SVMC) and Maximum Likelihood Classifier (MLC) were performed with a view to comparing the capabilities of SVMC and MLC using Post-processing Accuracy Assessment (PAA) and a Novel Template in producing urban land use and land cover map of the area. The objectives include performing supervised classification using SVM and MLC in ENVI Software, analysing the performance of SVM and MLC in mapping geometric features using error matrix and a new template. The methodology used comprise Image acquisition, Image enhancement, Image Sub-setting, Extraction of Regions of Interests (ROIs) and its separability index analysis, supervised classification using SVMC and MLC, Post-Processing Accuracy Assessment, Statistical Analyses, and Preparation of maps. ENVI 5.1 software was used for image processing, masking, spatial data analysis and classification. Microsoft Excel, GraphPad Prism ver.7.0 and IBM SPSS ver.21 were used for statistical analysis. The result of image classification indicates that Nnewi-North L.G.A is having 13.52% of Built-up Areas, 24.23% of Vegetation, 22.05% of Water bodies, Farm lands is equal to 39.40% and open/bare surface is 0.81% using SVMC while MLC result shows that Built-up Areas is14.99%, Vegetation is 13.01%, Water bodies is 34.08%, Farm lands is 36.00% and open/bare surface is 1.32%. Error Matrix and Kappa Coefficient results revealed that SVMC is better than MLC as follows (SVMC overall Accuracy is 98.07% and Kappa Coefficient is 0.97 while MLC overall Accuracy is 82.50% and Kappa Coefficient is 0.76. Additional statistical testing with aggregate mean from SVM and MLC was used to determine the significance of the mean difference using the researcher’s developed template called “Post Confusion Matrix” (PoCoMa). The result showed that the t-statistics is 0.670 with probability value of -0.476 which is greater than 0.05, thus, the null hypothesis was accepted with a deduction that using any of the algorithms (SVM and MLC) yields no significant difference in performance and efficiency of result of the map produced. The overall study revealed that both classifiers are efficient and accurate statistically, without any significant difference but using error matrix analysis, the research revealed that ‘Support Vector Machine Classifier’ is robust in extracting urban landscape from HRSI, especially Built-up areas and open/bare surfaces. The research recommends there is need for periodic urban LULC analysis to guide stakeholders in Planning, Monitoring, and Management of ‘Urban Areas’ among others. Keywords: Support Vector Machine Classifier (SVMC), Maximum Likelihood Classifier (MLC), Post Confusion Matrix (PoCoMa), High Resolution Satellite Image (HRSI), ENVI 5.1 software, and GraphPad Prism ver.7.0. DOI : 10.7176/JIEA/9-4-06 Publication date :June 30 th 2019
Highlights
In the recent decades, remote sensing imagery makes the monitoring of the earth surface and atmosphere possible
This tends to suggest that there could be a variation in the performance efficiency of map production and reading from Support Vector Machine Classifier (SVMC) and Maximum Likelihood Classifier (MLC)
It is worthy to note that the availability of various algorithms through machine learning approaches improves the capability of image classification in an intelligent way, the High Resolution Satellite Image (HRSI) GeoEye1, (2016) was used to produce urban land use and land cover map of Nnewi North Local Government Area using ENVI Support Vector Machine (SVM) classifier and MLC algorithms
Summary
Remote sensing imagery makes the monitoring of the earth surface and atmosphere possible. A number of classification approaches have been developed over the past decades and a review of these algorithms can be found in Lu and Weng (2007) In this regard, support vector machine, originally based on binary function, is viewed as one of the new ways of improving classification accuracies in remote sensing studies (Foody & Mathur, 2004a; Huang et al, 2007). Processing large images through the SVM classifier is time-consuming at high resolution, so ENVI's SVM provides a hierarchical, reduced-resolution classification process that improves performance without significantly degrading results It is most effective when operating in areas that contain homogenous features, such as water bodies, parking lots, and fields. The success of the SVM depends on how well the process is trained Regions of interests (ROIs) was used to train the HRSI and result was compared with that of maximum likelihood classifier
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
