An Ar Natural Marker Similarities Measurement Algorithm For E-Biodiversity
2018
In the last few decades, different techniques have been studied and proposed for flower species
classification. Nonetheless, the outcomes of these research are particular in term of the assessed
stages of classification conduit, the adopted data for assessments, and in the comparative baseline
methods. The objective of this research is to comparatively evaluate the effectiveness of different
algorithms, method combination procedure, and their parameters towards classification accuracy.
Algorithms of investigation starting with span from extraction, matching and classification to
determine the interest point of flower species, like colour and shape features information. This
research has been found out that the feature extraction process in Augmented Reality (AR) system
can be combined into Content-Based Image Retrieval (CBIR) system yield higher classification
results in efficiency and accuracy. The accurate identification of image features can reduce the
computational complexity, time consuming and enhance the accuracy of the identification and
classification for flower species. The proposed method can successfully reduce the number of
interest point by 89.98 percent. In addition, the computational complexity can be reduced from
O(n log n) to O(n), and the percentage of average accuracy for classifying flower species had
reached 98.8 percent.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
0
References
0
Citations
NaN
KQI