Pengembangan Metode Zoning untuk Pengenalan Pola Menggunakan Transformasi Slant

Abstract

In pattern recognition, feature extraction with zoning method is considered as one of the most effective methods for extracting characteristic of the pattern. However, the handwriting pattern recognition need another approach based zoning methods to get better results for handwriting pattern recognition. This research aims to propose a good handwriting pattern recognition with the development of methods of zoning and transformation slant. Development of methods of zoning is done by dividing the result by the feature extraction method of zoning partition into partitions 7 models, namely the partition lx4, 4xl partition, 2x2 partitions, lx8 partitions, 8xl partitions and 2x2 partitions. The result of seven models feature extraction on the partition reprocessed with slant transformation to produce a value called the energy stored in the database. Testing is done by matching the new patterns were tested with the energy patterns in a database whether it can be recognized or not. Results of tests performed, the level of pattern recognition on the zoning partition 4x4 are 90,89%, on the non zoning partition are 44,89% , on the zoning partition lx4 are 69,56%, on the zoning partition 4xl are 80,22%, on zoning partition 2x2 are 88%, on the zoning partition lx8 are 73,78% and on the zoning partition 8xl are 83,11%.