Authors : Tawfiq A. Al-Asadi and Fanar Ali Joda
Abstract: Object detection is the most important stage due to its great impact on next stages. Moving objects detection in dynamic environment has several challenges such as dynamic background, illumination changes (gradual, sudden), noise and etc. Traditional techniques of background modeling do not have the ability to face these challenges since, they can deal with a static background. Based on spectral, spatial and temporal features, we have proposed multi-modal method to model background and detecting moving object for video surveillance system it can deal efficiently with these challenges. The proposed system consists of two stages, the first one is construction of background model by select N frames to create group of histograms for each pixel. The second one is foreground detection and maintenance scheme for each pixel two histograms are constructed: histogram of ACS-LBP descriptor to extract local texture patterns and histogram of hue channel to extract color information. Temporal values fused within ACS-LBP histogram of intensity. Then maintenance scheme for background model is implemented by using two learning rate. Experiments on video challenging sequences illustrate the efficiency of the proposed method compared to the benchmark methods.
Tawfiq A. Al-Asadi and Fanar Ali Joda, 2018. Object Detection Based on Fusion of Multi-Feature for Video Surveillance System. Journal of Engineering and Applied Sciences, 13: 8159-8166.