ABSTRACT:
Moving object recognition system is an important step in any video tracking activity or surveillance system. In literature, moving shadows and clutters constitute a problem in moving object detection. Hence, the aim of this paper is to employ Adaptive Gaussian Mixture Model (AGMM) in recognizing human and vehicular motions in video images. The result of the AGMM using the Maximum A posterior (MAP) updates on video clips (dataset) obtained from Adeyemi College of Education Ondo, Nigeria showed a reliable (better slow start and illumination/light changes) detection accuracy. In most object detection algorithms, moving shadows/clutters can be mistaken as moving objects. Therefore, shadow/clutters were suppressed using a two-level shadow removal schemes, the HSV (Hue Saturation Value) and Phong illumination Models. The overall performance of this system was evaluated using the confusion matrix and the receiver operating characteristic (ROC), shadow detection and shadow discrimination values which showed a better result compared to existing benchmarks. Moreso, the sensitivity of this system was tested against the target presence using percentage of correct classification PCC and the Receiver Operating Characteristic ROC, which showed that the detection of moving objects (Human) has PCC of 96.28% and Area Under Curve AUC of 0.76. Similarly, Vehicle has PCC of 97.33% and AUC of 0.78; while the classification results in PCC and AUC of 98% and 0.8 respectively.
Keywords:
Moving object, Gaussian Mixture Model (GMM), ROC, Confusion Matrix, Evaluation.
