Learning Patterns of Activity Using Real-Time Tracking
IEEE Transactions on Pattern Analysis and Machine Intelligence
Analysis of the Clustering Properties of the Hilbert Space-Filling Curve
IEEE Transactions on Knowledge and Data Engineering
Advantages of Using a Space Filling Curve for Computing Wavelet Transforms of Road Traffic Images
ICIAP '03 Proceedings of the 12th International Conference on Image Analysis and Processing
A new motion detection algorithm based on Σ-Δ background estimation
Pattern Recognition Letters
Centre of mass model - A novel approach to background modelling for segmentation of moving objects
Image and Vision Computing
Motion detection using Fourier image reconstruction
Pattern Recognition Letters
Object tracking using SIFT features and mean shift
Computer Vision and Image Understanding
Complete-to-overcomplete discrete wavelet transforms: theory and applications
IEEE Transactions on Signal Processing
Statistical modeling of complex backgrounds for foreground object detection
IEEE Transactions on Image Processing
Determination of road traffic parameters based on 3d wavelet representation of an image sequence
ICCVG'12 Proceedings of the 2012 international conference on Computer Vision and Graphics
| Hi-index | 0.00 |
The detection of moving objects in video streams is generally performed by analysis of the differences between the modelled background and the current stream content, by matching object models, extracting and clustering the features of objects or else by using various filtering methods. Filtering is performed on the transformed contents of the video stream. Due to implementational constraints, mainly limited processing resources, solutions based on these principles of detection are sensitive to ambient light variations, objects shadows and camera movement. This paper presents a method for the detection of moving objects that uses a data reduction technique based on wavelets. Instead of the analysis of raw video data, wavelet coefficients of an appropriate scale are explored. In order to satisfy low processing requirements, an integer version of discrete wavelet transform is chosen for processing. To facilitate the detection, each frame is converted into a vector of pixel values. Consecutive video vectors are transformed using one-dimensional Discrete Wave Transform (DWT). The computed DWT coefficients make up a surface, which maps changes in their values over time. The surface is analysed to find clusters of values corresponding to moving objects. The checked patches represent moving objects. The width of a patch indicates the object size. Background details and illumination changes are represented by gradually changing patterns. Various examples demonstrate the potential of the method for practical applications.