| Autor: |
Prasad, C. H. Vaikunta Krishna, Amudha, V. |
| Předmět: |
|
| Zdroj: |
AIP Conference Proceedings; 2024, Vol. 2871 Issue 1, p1-8, 8p |
| Abstrakt: |
The study's overarching goal is to evaluate the efficacy of two different algorithms for SAR images categorization: the K-nearest neighbour technique and a new convolution neural network. We extracted 20 data instances from the Kaggle database system. The dataset is separated into two sets as training and testing. In 20 dataset 10 used for training and 10 used for testing. Classification of synthetic aperture radar pictures and comparison of K-nearest neighbours' and the unique convolution neural network's identification efficiency are conducted in MATLAB. The sampling number is determined using the G power formula, with the variables alpha set at 0.05 as well as power at 80%. From the MATLAB simulation the novel convolutional neural network algorithm obtained a classification accuracy of 95.31% and K-nearest neighbour algorithm obtained classification accuracy of 73.01%. Utilizing SPSS evaluation, we found a significance value of 0.005 (p<0.05). In conclusion, the new convolution neural networks outperform the K-nearest neighbour method in terms of effectiveness on the given dataset for artificial aperture radar picture classification. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
