Identification of Lane Lines Using Advanced Machine Learning

The automobile field is getting smarter and more robust in day today life. Lines on lanes represent the rules for mobile of vehicles. Identification of these lane lines is a challenging and requires advanced concepts in the area of computer vision. The resultant of this work may be applicable for auto piolet engine vehicles. This manuscript represents the lane line segmentation using watershed segmentation algorithm, JRip and Random Forest machine learning algorithms. The readily captured colored images treated as input images and instead of markers-based segmentation, morphology-based segmentation is performed using a watershed segmentation algorithm. After segmentation, the structural similarity index(SSIM) results in an average of 99.89%. Gray level co-occurrence matrix (GLCM) method is adopted for extracting six different features from the image and classified using JRIP classifier which achieves 97% of accuracy and Random Forest classifier achieves 93.5% accuracy. The JRip classifier obtains good accuracy results when compared to the random forest classifier. The entire work is performed using python language with project Interpreter of Python 3.8.