| 主权项 |
1. A method for detecting traffic video information, comprising:
acquiring a traffic video stream; determining color features of each frame of image in the traffic video stream; calculating inter-frame distance between adjacent frames according to the color features; calculating boundary of an image clustered frames' group according to the inter-frame distance by adopting an image clustering evaluation standard in RGB space and an image clustering evaluation standard in YUV space respectively; and determining a final boundary of the image clustered frames' group according to boundaries of the image clustered frames' group in RGB space and YUV space, wherein the determining the color features of each frame of image in the traffic video stream comprises: sampling each frame of image to obtain sampled data; generating histograms of R, G, B and Y (brightness or luminance) color channels of each frame of image according to the sampled data; and expressing the color features of each frame of image by using four-dimensional histograms as F={HRi, HGi, HBi, HYi}, wherein HRi, HGi, HBi, HYi, are histograms of four color channels of R, G, B and Y respectively, i=1,2, . . . , N and N is number of bins of each divided color channel, wherein the calculating the inter-frame distance between the adjacent frames according to the color features comprises: calculating the inter-frame distance of each color channel between the adjacent frames in RGB space; finding an intersection of inter-frame distances corresponding to three color channels of R, G and B; obtaining the inter-frame distance in RGB space by determining the minimal and maximal inter-frame distances of the three color channels of R, G and B respectively; and obtaining the inter-frame distance in YUV space by calculating the inter-frame distance of Y channel between the adjacent frames. |
