Wood species recognition based on phase - only correlation (POC) technique
This research was performed in order to determine suitable methodology and techniques to automatically. identify wood species. Traditional wood recognition requires employees who are skilled in recognizing wood species. The two main problems with recognition is the lack of skilled employees and the...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2013
|
| Subjects: | |
| Online Access: | http://umpir.ump.edu.my/id/eprint/37233/1/Wood%20species%20recognition%20based%20on%20phase%20-%20only%20correlation%20%28POC%29%20technique.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This research was performed in order to determine suitable methodology and techniques to automatically. identify wood species. Traditional wood recognition requires employees who are skilled in recognizing wood species. The two main problems with recognition is the lack of skilled employees and the visually impaired employees. This leads to an interruption of the wood recognition process. Besides that, based on the previous research, two techniques are used to recognize wood species. These techniques often do not give a high accuracy of wood species recognition. Accordingly, to improve the accuracy rate of wood species recognition, this research was proposed. There are four main steps performed to develop a wood species recognition system. These are; data acquisition, pre- processing, features extraction, and features analysis. The technique used was the Phase-Only Correlation (POC) technique. The POC technique involves three steps during the feature extraction phase. These steps are the 2D Discrete Fourier Transform (2D DFT), Cross Phase Spectrum, and 2D Inverse Fourier Transform (2D IFT). Each step contains specific calculations and algorithms to obtain details on the wood’s features and match them with the possible matching species. By applying all three steps, the test image is found to either match or not match the training data. Next, the classification process is performed with the help of the K-Neighbour Network (KNN) technique; where it helps to classify the wood images. This technique was used to test eight wood species , namely Balau, Ramin, Durian, Melunak, Mersawa Gajah, Rengas, and Sepetir. From this testing, it was found that Balau and Ramin wood species were identified with 100% accuracy. Meanwhile, Melunak and Rengas wood species were identified to approximately 98.5%; Mersawa to 95%.%; and for both Durian and Sepetir wood species, 93.75% accuracy. was obtained. Furthermore, for the Mersawa Gajah wood species, only 91.25% eaccuracy was obtained. These research results show that 96.4% of the wood species tested accurately and were recognized without error. |
|---|