This repository contains the codes & results concerning the detection of lanes using BCD and Spline fitting
List of papers:
- Blind Color Decomposition
- Road detection from single image using texture orientation and confidence voting - well structured in introduction
- Color-based road detection plus confidence mapping
- HSV-based lane detection - results clearly shown
- fuzzy-based rural road detection - looks a good paper for result comparison on some images
- k-means based clustering + memory of roads - the C part of section should be integrated into our model -> here they use k-means clustering but EM uses k-means for initialisation of gaussain estimate and then iterates to improve the parameters, therefore better
- good previous work section here
Some to Dos for results and possible modifications:
- Check for speed of processing and use EM::train_startWithE after first frame using the means and covariances of previous training
- Try and see what happens if we use EM in susequent frames to predict road pixels using the area predicted as road in the previous image