.png){kind=icon}
.png){kind=icon}
Return
Lane Detection for Self-Driving Cars Using Mouse Customized Golden Eagle Optimizer (Mcgeo) and Two-Fold-Deep-Learning Model
.png){kind=icon}
PRE
AI
Q
V
Z
DOI：10.1142/S0218001425530027.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Road boundary lane detection has emerged as an essential task toward avoiding road accidents that occur predominantly because of vehicular unintentional crossing of road lane boundaries. Significant advances have been made so far in various Machine Learning (ML) algorithms meant for lane detection; however, it is indeed difficult to meet accuracy and efficiency levels. Thus, this work attempts to explore a new kind of driver alert related to cross-boundary lanes. The designed approach can be divided into four major steps, which are as follows: (1) preprocesses raw images by applying Sobel filters and Histogram Equalization; (2) performs foreground region identification using an optimized Fuzzy C-Means Clustering technique with the membership function of the clustering model being optimized by the hybrid optimization technique Golden Eagle Optimizer-Cat-and-Mouse-based Optimization (GEO-CMBO); (3) performs key feature extractions including color moments, edges, lines, and corners detections employing various image processing techniques like Prewitt edge detector, Probabilistic Hough Transform, and Harris corner detection; and (4) does feature selection using MCGEO followed by classification using Transformer Neural Network (TNN) and Recurrent Neural Networks (RNNs). The proposed system is implemented in MATLAB and tested based on various metrics such as Accuracy (99.7%), Sensitivity (99.9%), Precision (99.8%), Specificity (99.5%), False Positive Rate (FPR) (0.1%), and False Negative Rate (FNR) (0.2%). Comparative results show the superiority of the proposed method over existing models, thus having a high potential for real-world applications, especially in autonomous vehicles.
Keywords:
Lane detection
machine learning
fuzzy C-means clustering
hybrid optimization
autonomous vehicles
Journal
IF:
1.1
Papers:
161
Citations:
2.0K
Organization
Cited Papers