AbstractsComputer Science

Computer aided plant identification through leaf recognition using enhanced image processing and machine learning algorithms; -

by N Valliammal




Institution: Avinashilingam Deemed University For Women
Department: Computer Science
Year: 2014
Keywords: Sigmoid function, Genetic algorithm, Kmeans clustering
Record ID: 1216344
Full text PDF: http://shodhganga.inflibnet.ac.in/handle/10603/29171


Abstract

Computer aided identification of plants is an area of research that has gained more newlineattentionin recent years and is proving to be a very important tool in many areas including agriculture forestry and pharmacological science A general process of a Computer Aided Plant Classification through Leaf Recognition CAPLR contains four steps namely leaf image enhancement leaf segmentation feature extraction and classification The first step in CAPLR enhances a leaf image newlineby using an approach that simultaneously removes noise adjusts contrast and enhances boundaries newlineThe second stepuses a wavelet based segmentation approach that combine clustering with texture newlinebased color features to extract the leaf from its background A total of 28 features were extracted newlinewhich were grouped into five categories namely geometric features color features texture features newlinefractal features and leaf features To enhance the process of leaf recognition a fusion method is newlineproposed which combines Genetic Algorithm GA and Kernel Principal Component Analysis newlineKPCA with shared and merger operations in the third step The single and fused feature sets are then newlineused by classifier to recognize the leaves and identify the plants For this purpose a two level newlineclassification model was used where the first level classifier was used to produce an refined training newlineset which was used to train the second level classifier Two leaf image datasets namely standard and newlinereal were used during experiments that evaluated the performance of the proposed algorithms The newlineexperimental results showed that the two level classification algorithm improved the efficiency of newlinerecognition and identification in terms of accuracy and speed The various results showed that the newlinemodel WNN for the first classifier and SVM for the second classifier that used GA KPCA with leaf newlineand fractal produced high recognition rate newline%%%-