International Journal of Computer Applications |
Foundation of Computer Science (FCS), NY, USA |
Volume 175 - Number 18 |
Year of Publication: 2020 |
Authors: Khoerul Anwar, Sigit Setyowibowo, Sujito |
![]() |
Khoerul Anwar, Sigit Setyowibowo, Sujito . Voting Block Method for Verification of Beef and Pork using Back Propagation Learning Machines. International Journal of Computer Applications. 175, 18 ( Sep 2020), 33-37. DOI=10.5120/ijca2020920702
Identifying two digital image objects that have visually similar textures in the field of computer vision is a tough task. Usually the visual texture of an object has unique features from one another. However, the beef texture and pork texture both have similar visual textures. This visual similarity is often used by certain individuals to commit fraud in selling meat. It is important to conduct research on the texture similarity phenomenon in order to obtain a computational method that can be used to distinguish the two. So that this method can be applied in the field of computer vision. Until now, a mobile-based application for identifying beef and pork based on texture features in the community does not exist. In this condition, beef consumers often mistakenly recognize beef and pork. In this scientific paper, a beef and pork identification method is proposed based on the feature texture computation using the Vbloc method. The Vbloc (voting bloc) method is a classification technique that starts with partitioning the original image into n new image partitions (bloc). This bloc image is used as a testing image on ANN. It is different from the general classification where the training image and the testing image have the same dimensions. In the Vbloc method, the test image is cut into parts of a new image with smaller dimensions then the pieces are extracted. Next, testing is done on ANN to obtain the class from the input image. The method offered was tested on 61 pork images, 81 beef images and 10. The results obtained by the vblock method were able to identify the type of image correctly at 76.97%.