c++ - Lane detection in a most-shadowed image with some light -


currently i'm working on finding lane lines in image opencv , c++.

however, encountered big problem utilizing threshold on following picture.

original image

i have tried using thresholding methods below, none of them worked.

  1. sobel x filter min threshold 10 , max threshold 60
  2. sobel y filter min threshold 10 , max threshold 60
  3. r channel in bgr color space min threshold 30 , max threshold 60
  4. s channel in hls color space , u channel in yuv color space. no matter threshold took in these 2 color spaces, looks disaster.

take bgr-r example, lane lines clear bright part obvious.

i have tried histogram equalization balance white , black part in image before thresholding, enhanced contrast, makes result worse.

i don't know whether can categorize problem shadow-removal, because part in image covered shadow.

i wish can teach me pre-process steps take or possible methods can find line.

thanks in advance!


Comments

Popular posts from this blog

android - InAppBilling registering BroadcastReceiver in AndroidManifest -

python Tkinter Capturing keyboard events save as one single string -

sql server - Why does Linq-to-SQL add unnecessary COUNT()? -