computer vision - Illuminant variant image in road recognition -
recently i've been doing study on topic road recognition. there problem if have illuminant variant images, recognition results not satisfactory if use normal segmentation approach. here, shadow removal improve results lot, image, part of taken under strong lighting condition, part of road hard recognize, in picture in lower-left corner:

could please suggest approaches of solving problem?
when image bright (or dark) try contrast enhancement.
edit looking @ example image looks features washed out , contrast enhancement not detect white line marking edge of road. alternative might find line using line detector, , extend bottom of image. assume line continues same path.
if doing road recognition i'm guessing have (or will) implement method lane tracking on dotted lines in road. same logic used dotted line case applied image. bright area seen dotted line, algorithm should still know same lane
Comments
Post a Comment