Development of feature descriptors and a classifier for the classification of coral reef and coral reef diseases

Abstract

Coral reefs are the richest and the most diverse ecosystem found in the global ocean but is also among the most threatened Coral reef is a source of food for millions of marine species It protects astlines from storms and erosion It is a source of new medicines for treating human bacterial infections cancer viruses arthritis and heart diseases It also assists in carbon and nitrogen fixing as well as helping with nutrient recycling Nowadays coral reef assification is an emerging concept in image processing and machine learning The main objective of this research work is to develop novel feature descriptors that aid in achieving high classification accuracy for classifying coral reef images coral reef diseased images and videos To realize this objective five feature descriptors namely Improved Local Derivative Pattern ILDP Z with Tilted Z Local Binary Pattern Z TZLBP Octa angled Pattern for Triangular sub region OPT Mean Direct Code Pattern MDCP and Diagonal Direction Value Pattern DDVP are proposed in this research work The first feature descriptor proposed is an improvement to LDP and termed as ILDP which has reduced the bin size of the histogram thereby reducing the time complexity involved improving accuracy ILDP template extracts local information by encoding various spatial relationships that capture detailed information. newline