Comparison of SVM And BPNN Methods in The Classification of Batik Patterns Based on Color Histograms And Invariant Moments

Currently, various types and patterns of batik in Indonesia have been widely documented. Nevertheless, there are many obstacles in classifying batik's data, because the classification of batik has not been based on standard types of motifs. In this research, classification of batik's motif patterns based on color histograms and invariant moments was carried out. For the batik's pattern recognition method, two methods are tested and compared, namely the Backpropagation Neural Network (BPNN) method, and the Support Vector Machine (SVM) method. The speed of the pattern recognition process of batiks' motif using SVM method requires an average processing time of 0.77 milliseconds, while the BPNN method requires an average time of 3.59 milliseconds. The accuracy of the classification of batik's pattern test data using the SVM method has an accuracy of 88.33 percent, while the BPNN method has an accuracy of 76.25 percent. The development of this research can be done by improving methods and variations in image data of batik's motif pattern.