PENGENALAN POLA RAMBU LALU LINTAS MENGGUNAKAN ALGORITMA PRINCIPAL COMPONENT ANALYSIS DAN HISTOGRAM OF ORIENTED GRADIENT SEBAGAI EKSTRAKSI FITUR CITRA

Tontowi Prasetyo, NIM. 13650050 (2018) PENGENALAN POLA RAMBU LALU LINTAS MENGGUNAKAN ALGORITMA PRINCIPAL COMPONENT ANALYSIS DAN HISTOGRAM OF ORIENTED GRADIENT SEBAGAI EKSTRAKSI FITUR CITRA. Skripsi thesis, UIN SUNAN KALIJAGA YOGYAKARTA.

Preview	Text (PENGENALAN POLA RAMBU LALU LINTAS MENGGUNAKAN ALGORITMA PRINCIPAL COMPONENT ANALYSIS DAN HISTOGRAM OF ORIENTED GRADIENT SEBAGAI EKSTRAKSI FITUR CITRA) 13650050_Bab I_Bab Terakhir_Daftar Pustaka.pdf - Published Version Download (5MB) | Preview
	Text (PENGENALAN POLA RAMBU LALU LINTAS MENGGUNAKAN ALGORITMA PRINCIPAL COMPONENT ANALYSIS DAN HISTOGRAM OF ORIENTED GRADIENT SEBAGAI EKSTRAKSI FITUR CITRA) 13650050_Bab II_SD_SEBELUM_BAB_TERAKHIR.pdf - Published Version Restricted to Registered users only Download (9MB)

Abstract

Research on the traffic sign recognition begin recently in line with the growing popularity of intelligent vehicle systems. Traffic sign recognition has its own challenges which include pre-processing, feature extraction, also detection and classification processes. In addition, the condition of signs when images are taken such as lighting conditions and image clarity can also affect the classification process. The purpose of conducting this research is to support the system of smart driver and car assistants. This research aims to analyze the process of introducing traffic signs using the Principal Component Analysis algorithm and Histogram of Oriented Gradient as feature extraction. Other parameters in the introduction process are also used, such as pre-processing and classification using artificial neural networks.The testing scenario includes three pre-processing parameters including: first parameter - Median Blur, Grayscale, Histogram Equalization, Binary, and Canny; second parameter - Graysale, Histogram Equalization, Binary, and Canny; and the third parameter is using no pre-processing. Whereas, for neural network structure using four hidden layer parameters: 10, 50, 100, and 150. The training process is carried out by combining predetermined scenarios. The training results are grouped into two major parts based on feature extraction. The Histogram of Oriented Gradient Algorithm has a high accuracy compared to the Principal Component Analysis algorithm with a maximum accuracy of 100%. While the Principal Component Analysis algorithm has a maximum accuracy of 79.38%. In the testing process, the Histogram of Oriented Gradient algorithm has an accuracy of 80.84%, while the Principal Component Analysis algorithm only has the highest accuracy of 23.79%.

Share this knowledge with your friends :

Actions (login required)

View Item