Free Space Detection for autonomous navigation in daytime

Free Space Detection for autonomous navigation in daytime foggy weather Nicolas Hautière, Jean-Philippe Tarel, Didier Aubert

Light under Daytime Fog Daylight Atmospheric veil Scattering Direct transmission 2

Light attenuation by the atmosphere § Koschmieder’s law: Apparent luminance Object luminance Extinction coefficient Object distance Atmospheric luminance 3

Visibility Range under Daytime Fog Visibility distance: “the greatest distance at which a black object of suitable dimensions can be recognized by day against the horizon sky” (CIE, 1987) § From the Koschmieder’s law ( express the contrast of an object against the sky: Þ ) let contrast Attenuation § For a black object (C 0=1) and a visibility contrast threshold of 5%: 4

Flat road assumption n Assuming a flat road, the depth of a road point is: C x f z v y q u vh Image plane H where: n n n vh the horizon line the pixel size. S X Y Z Road plane M d 5

Exploitation of the atmospheric veil Method: instanciation of the Koschmieder’s law Assuming that the camera response function is linear, the Koschmieder’s law becomes within the image space: B&W image Extraction of a region of interest Fitting of a measurement bandwidth Vmet = 50 m b estimation thanks to the inflection point vi: Measurement and derivation of intensity curve Extraction of the inflection point Estimation of the meteorological visibility distance [Hautière et al. , 2006 a] Hautière, N. , Tarel, J. -P, Lavenant, J. and Aubert, D. (2006). Automatic Fog Detection and Measurement of the visibility Distance through use of an Onboard Camera. Machine Vision Applications Journal, 17(1): 8 -20 6

Recovery of the object luminance (1) § Extinction coefficient b is now determined § The Atmospheric luminance A∞ is given by the bandwidth above the horizon line § Lets compute R by reversing the Koschmieder’s law: § There is still one unknown: d 7

Recovery of the object luminance (2) § The previous equation may be rewriting as follows: § The contrast after restoration with respect to the background sky is thus: § The contrast restoration is exponential =0. 05 (visibility = 60 m), A =255 8

Free space segmentation § By using a flat world assumption the vertical objects are falsely restored (their distance being largely overestimated) § Their intensity becomes null after the restoration process § This drawback may be used in our advantage to segment the vertical object § The free space is thus segmented by looking for the biggest connected component in front of the vehicle (whatever the method). 9

Free space segmentation (Foggy weather) 10

Free space segmentation (rainy weather) 11

Thank you for your attention 12