I am working on the texture feature extraction from foot for person identification. I applied Gray Level Co-occurrence Matrix, it provides the accuracy of 87%. What are the factor which can be modified for getting good accuracy.
Dear Riti Kushwaha,
If you would like to get accurate feature extraction, multilevel discrete wavelet transformation is the perfect choice. It can analyse the message for multi-frequency bands, so you can use some statistical calculation such as mean, variance, etc to get different features for different bands. Also, I recommend to apply Window function such as Hamming window on the selected bands before statistical analysis.
For more information follow the links:
https://pdfs.semanticscholar.org/0995/388b96f50d0ef67edd3c4a09b8d79cd01e45.pdf
https://pdfs.semanticscholar.org/29d4/2c2c4c23cf3ed4145ef08c57caec301155e8.pdf
https://ieeexplore.ieee.org/document/920996
Thank you for the suggestion. I will try with discrete wavelet transform.
- Badges
- Science topic
- Similar topics
- Mathematics
- Algebra
- Matrix