PERBANDINGAN JARAK POTRET DAN RESOLUSI KAMERA PADA TINGKAT AKURASI PENGENALAN ANGKA KWH METERMENGGUNAKAN SVM

Abstrak

Electricity meter is a tool used to measure the electricity power consumption. Electricity meter has the numeral part known as electricity meter number which load the electricity power consumption.Usually, the recording is done once a month, by recording the number on the electric meter into the book, and then the recording officer take pictures the power meter. Electricity meter number can be analized through an image using the knowledge of pattern recognition in image processing. In analyzing picture of electicity meter number, we used Histogram Of Oriented Gradients (HOG) as the feature extraction and Supply Vector Machine (SVM) as the classification method. The result using 100 train-set and 30 test-set for each combination of category shows that the best resolution is 10 MP and 14 MP and the picture capturing distance is at 30 cm byand 10 cm by 73,33%  accuracy for each image and 86,67% for each number and confusion matrix shows that presentation of all number is 75,48%.

 Key words:Number Recognition, Histogram Of Oriented Gradients (HOG) , Support Vector Machine (SVM)