DETEKSI RAMBU LALU LINTAS BERBASIS SEGMENTASI WARNA ADABOOST DAN K-NN
Daftar Isi:
- Traffic sign detection automatically can help motor vehicle drivers to recognize and obey the traffic sign. However, many things affect the traffic sign not to be detected, such as high light intensity in daytime or low in the evening. Therefore, it needs a software that can detect traffic sign in those conditions. In this research it has been developed a software that could detect traffic sign, either in daytime or evening with AdaBoost color segmentation and k-NN method. Weak learner in AdaBoost was trained to perform the segmentation to image based on the color of traffic sign which would be detected. While k-NN was used to verify by comparing the object which was marked before using histogram projection, and the saved template images. The experiment was conducted for 3 times on 40 test data, with the number of AdaBoost training data which was smaller, equal to and greater than the test data. Each of these experiments resulted the accuracy value of 62.5%, 65.0% and 100.0%.