Sequential Operations in Digital Picture Processing
Journal of the ACM (JACM)
Non-parametric Model for Background Subtraction
ECCV '00 Proceedings of the 6th European Conference on Computer Vision-Part II
Efficient adaptive density estimation per image pixel for the task of background subtraction
Pattern Recognition Letters
A Convolution Kernel Method for Color Recognition
ALPIT '07 Proceedings of the Sixth International Conference on Advanced Language Processing and Web Information Technology (ALPIT 2007)
Deciding the Number of Color Histogram Bins for Vehicle Color Recognition
APSCC '08 Proceedings of the 2008 IEEE Asia-Pacific Services Computing Conference
Image segmentation in video sequences: a probabilistic approach
UAI'97 Proceedings of the Thirteenth conference on Uncertainty in artificial intelligence
The efficient algorithms for achieving Euclidean distance transformation
IEEE Transactions on Image Processing
Vehicle Detection Using Normalized Color and Edge Map
IEEE Transactions on Image Processing
A view-invariant and anti-reflection algorithm for car body extraction and color classification
Multimedia Tools and Applications
Hi-index | 0.00 |
The vehicle color classification (VCC) becomes a challenge problem in the video surveillance system since the foreground mask derived by the model-based background subtraction algorithm will mislead the VCC by the impure background and the improper foreground regions. In this paper, a novel VCC algorithm based on refining the foreground mask with following two steps is presented to ensure classification result. First, by combining the results of model-based foreground mask and image segmentation, we translate an image into several regions based on the newly developed mask-based connected component labeling algorithm. Secondly, the refined foreground mask is derived by analyzing the region property to remove the undesired regions. The support vector machine (SVM) based classifier with two-layer structure is adopted in this paper. The first layer classifies the image into color and grayscale classes, and the second layer contains two SVM classifiers for the two classes, respectively, which further determines the probability of 7 colors (black, gray, white, red, yellow, green and blue) of the object. It found the average correct rates are respectively 30.2% and 72.3% for the original and refined foreground masks that shows the proposed approach is promising to automatic determination of vehicle colors in a video surveillance system.