Journal of Applied Computer Science and Technology (Jun 2022)

Metode Otsu dan Mathematical Morphology Dalam Segmentasi Region Karakter Plat Nomor Kendaraan

  • Yovi Apridiansyah,
  • Rozali Toyib,
  • Ardi Wijaya

DOI
https://doi.org/10.52158/jacost.v3i1.277
Journal volume & issue
Vol. 3, no. 1
pp. 134 – 143

Abstract

Read online

The problem that affects the character segmentation step is the step before character segmentation, namely preprocessing character segmentation or called preprocessing. This poses are strongly influenced by plate lighting conditions, shadows against plates, plate image impurities, plate image resolution, character cutting accuracy, and speed in recognizing characters. In general, identification consists of 3 stages, namely detection, segmentation and recognition. In this study, the use of the otsu method is expected to detect the region on the vehicle license plate, the region in question is the first region to show the regional code, the second region for the registration number and the third region for the sub-region code. In the process, the results of the vehicle number plate detection trial to get a segmentation of 3 regions of the vehicle number plate character did not get the expected results. The results trial obtained the identification of the entire character of the vehicle number plate so that the characters on the vehicle number plate could not be distinguished between the front letter, number, and the back letter. So to maximize the desired results so that getting 3 regions of the otsu method segmentation process needs to be improved using the mathematical morphology method. This mathematical morphology method serves to read the character value of each pixel in the digital image which produces a comparison between the pixels in the image, so morphology techniques are appropriate when used to perform image processing in obtaining the region of the vehicle number plate. From the improvement of the otsu method, the results of the trial were improved. Of the 100 data samples tested, 96 data samples passed and 4 sample data failed, the accuracy value using MSE measurements from the tested data samples received a very high increase, which was 96%.

Keywords