Study on method of detecting preceding vehicle Jilin university, CHINA 2004 IEEE CNF
流程圖 Detecting possible object Measure of symmetry Distance Detction
Character of Target Vehicle Gray characteristics Boundary characteristics Symmetry characteristics
Detecting possible object The triangle area is called the primary area of interesting
Detecting possible object G(r) is the mean value of the pixel gray on r row rb(r) is the coordinate of pixel on right end of r row in the primary area of interesting lb(r) is the coordinate of pixel on left end of r row in the primary area of interesting g(r,c) is a gray of pixel(r,c) on r row
Detecting possible object
P(x k ) is probability of the gray value x k
Detecting possible object
Identifying area of interesting
Measure of symmetry w is the width of symmetry area x s is origin u = x-x s
Measure of symmetry Energy function The Energy function of even function and odd function
Measure of symmetry Transform the even function let its mean value is also zero.
Measure of symmetry The range of symmetry measure is (-1,1) s=1, it is a full symmetry s=-1, it is no symmetry The symmetry depends on parameters x s and w
Measure of symmetry
Calculate the symmetry measure about every row in the identifying area of interesting. Count the mean value of symmetry measure.
Distance Detction f is the focus of CCD camera. αis the incline angle of CCD camera h is the height of CCD camera to ground (x,y) is the point P projection in image plane
Distance Detction image plane coordinate : millimeter, (x,y) store frame coordinate : pixel, (u,v)
Experiment for detecting preceding vehicle No Alg : overall detection rate 90% false alarm rate 13% mean time of detecting object 82ms Have Alg : verall detection rate 94% false alarm rate 6% mean time of detecting object 68ms
Experiment for distance detection