Most of background subtraction methods represent background statistics using probabilistic unified frame-works such as the Gaussian mixture model or kernel density estimation. But these models cannot define the exact difference between two pixels. It causes misclassi-fication such as false alarms and misses. We presented a new sensor noise model appropriate for general CCD cameras. Based on this, we propose a novel background subtraction method. Our noise modeling needs a line estimation step to relate image intensities with parame-ters of the noise distribution. This paper describes a new line estimation algorithm given two consecutive static images, and from which can have a well-fitted distribu-tion for each pixel according to intensity of the pixel. In addition, we present a background update method to deal with the continuous variation of the background. We can estimate accurate foregrounds by adapting the esti-mated per-pixel distributions and background updates.
조회 수 1870 댓글 0
|저 자||Youngbae Hwang, Hanbyul Joo, Jun-sik Kim, In So Kweon|
|학 회||IAPR workshop on Machine Vision Applications (MVA)|
Prev UAV Attitude Estimation by Combining Horizon-based and Homography-based Approaches for Catadioptric Images UAV Attitude Estimation by Combining Horizon-based and Homography-based Approaches for Catadioptric Images 2009.09.18by 김원진 Stereo Matching with the Distinctive Similarity Measure Next Stereo Matching with the Distinctive Similarity Measure 2009.09.16by