LAMPUNG HANDWRITTEN CHARACTER RECOGNITION USING HISTOGRAM OF ORIENTED GRADIENT
Recognition of Lampung handwritten characters has achieved an accuracy value of 94.50% by the latest research. The features used in previous research are a combination of features, namely BED (branch, end, and density) and water reservoir. With the number of features combined, the recognition proces...
Saved in:
| Main Author: | |
|---|---|
| Format: | Theses |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/36689 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Recognition of Lampung handwritten characters has achieved an accuracy value of 94.50% by the latest research. The features used in previous research are a combination of features, namely BED (branch, end, and density) and water reservoir. With the number of features combined, the recognition process also requires different algorithms at the same time. It is necessary to analyze to find the best features in recognizing Lampung handwritten characters..
The Histogram of Oriented Gradient (HOG) feature can describe the character's writing based on direction and gradient. In this study the best parameter selection has been carried out on the HOG feature for character recognition of Lampung characters. Classification with multi layer perceptron using the HOG feature, has been able to increase accuracy in Lampung handwritter characters recognition. We have tested the model using new data, this process is to measure the robustness of the model from outer data.
Testing the results of the introduction of Lampung characters with the HOG feature with the MLP classification succeeded in increasing the best accuracy to 99.32%. Model testing of new data successfully showed robustness performance of 99.58%. |
|---|