Automatic Detection and Classification of Road, Car, and Pedestrian Using Binocular Cameras in Traffic Scenes with a Common Framework
Algorithm 1
Road peripheral enveloping algorithm.
Require: : Disparity image, ,: Disparity values corresponding to the maximum and minimum intensity points for each line, : Disparity value of the th row in the V view, : Disparity threshold, : Euclidean distance, : Represents the cumulative sum of the Euclidean distances of the disparity values of adjacent rows in the V view;
Ensure: Final V view;
(1) for e doach frame
(2) Original performs horizontal and vertical projection to obtain initial U and V views;
(3) Remove large obstacles in the initial U view;
(4) The new U view is inversely transformed into a new disparity image and a new V view is obtained. The disparity value range of each row of the new V view is ;
(5) Whiledo
(6) The new V view is scanned progressively from bottom to top, removing disparity value points that do not meet the condition;
(7) if n thnew V view still has noise
(8) Find the line where the noise is located, make the point where the disparity point of this line meets as the reference point of the road, and update the range of road deviation where the noise is located to ;
(9) end if
(10) Obtain the Final V view using the considerate U-V view method.
