A Robust Moving Object Detection Approach
CISP '08 Proceedings of the 2008 Congress on Image and Signal Processing, Vol. 4 - Volume 04
Motion-based background subtraction using adaptive kernel density estimation
CVPR'04 Proceedings of the 2004 IEEE computer society conference on Computer vision and pattern recognition
| Hi-index | 0.00 |
An efficient ghost removal technique is proposed as an extension to adaptive background differentiation for motion detection. The pixels of the first frame in the sequence representing moving objects are replaced with the values taken for the same pixels from the memory bank where those pixels are identified as non-moving ones. The memory bank is built of the frames immediately following or, alternatively preceding, the initial frame of the analyzed sequence. This allows creating the initial background model with no moving pixels. Parameters optimization is conducted for specific case of traffic control system application. Experiments demonstrate that threshold reduction is beneficial to achieve completeness of the ghost removal. Additionally, a second improvement is introduced to reduce for the noise by non-stationary cameras which are shown to be efficiently compensated by a second derivative in the temporal differentiation when working with videos at a sufficiently high frame rate.