We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
DETECTION OF EDIBILITY OF AMLA (Emblica officinalis) THROUGH PCA BASED IMAGE ANALYSIS.
- Authors
SARKAR, Tanmay; MUKHERJEE, Alok; CHATTERJEE, Kingshuk; CHOUDHURY, Tanupriya
- Abstract
Identification of edibility of fruit samples is very essential, as well as difficult. This is more applicable in places where bulk fruits are used in different automated factories, where investigation of each fruit manually is an impossible task. In this work, we have proposed a principal component analysis (PCA) based threshold classifier scheme for the identification of edibility of amla fruits. We have analyzed only the hue histogram of the image samples using PCA to segregate the samples into Good, Intermediate or Bad classes. Use of analysis like PCA reduces the computational burden many folds compared to the other supervised learning models involving variants of neural network, or mathematically heavier transform based models like wavelet or Fourier transforms. The model is validated using different sample images. High accuracy of 98.33% for classification of samples is achieved in this work. Most importantly, low computational burden and non requirement of any other pre-filtering method to the sample images highlights the effectiveness of this algorithm.
- Subjects
IMAGE analysis; SUPERVISED learning; PRINCIPAL components analysis; WAVELET transforms; FOURIER transforms
- Publication
Economic Computation & Economic Cybernetics Studies & Research, 2022, Vol 56, Issue 2, p77
- ISSN
0424-267X
- Publication type
Article
- DOI
10.24818/18423264/56.2.22.06