Open AccessSemantic Labeling of natural Scene Images Using Color Features
Author(s) -
Kyawt Kyawt Htay,
G. R. Sinha,
Hanni Htun,
Nwe Ni Kyaw
Publication year - 2019
Publication title -
csvtu international journal of biotechnology, bioinformatics and biomedical
Language(s) - English
Resource type - Journals
ISSN - 2455-5762
DOI - 10.30732/ijbbb.20190402006
Subject(s) - artificial intelligence , computer science , computer vision , hsl and hsv , color histogram , pattern recognition (psychology) , histogram , color image , classifier (uml) , local color , color space , color quantization , color normalization , feature (linguistics) , image (mathematics) , image processing , virus , virology , biology , linguistics , philosophy
Scene image classification systems firstly need to locate the objects, and then classify the whole image. The color feature is importance to describe the properties of an image surface. The paper presents a framework for scene images to label local regions using color features. The paper uses maker-controlled watershed algorithm to segment the input image into regions. This paper uses the segmented regions as a basic input unit, and then extract Color Histogram (CH) and Color Moment (CM) features in HSV space. This system performs labeling using 3-layer Feed Forward Neural Network (FFNN) classifier. The system tests accuracy on public Microsoft Research Cambridge (MSRC) 9-class dataset.