Car Surveillance Video Summarization Based On Car Plate Detection

Today, video is a common medium for sharing information. Navigating the internet to download a certain form of video, it takes a long time, a lot of bandwidth, and a lot of disk space. Since sending video over the internet is too costly, therefore video summarization has become a critical technology. Monitoring vehicles of people from a security and traffic perspective is a major issue. This monitoring depends on the identification of the license plate of vehicles. The proposed system includes training and testing stages. Training stage comprises: video preprocessing, Viola-Jones training, and Support Vector Machine (SVM) optimization. Testing stage contains: test video preprocessing, car plate (detection, cropping, resizing, and grouping detecting test car plate, feature extraction using HOG feature. The total time of local recorded videos is (19.5 minutes), (15.5 minutes) for training, and (4 minutes) for testing. This means, (79.5%) for training and (20.5%) for testing. The proposed video summarization has got maximum accuracy of (86%) by using Viola-Jones and SVM by reducing the number of original video frames from (7077) frames to (1200) frames. The accuracy of the Viola-Jones object detection process for training 700 images is (97%). The accuracy of the SVM classifier is (99.6%).

Temporally Consistent Superpixels

This thesis addresses the field of early stage video preprocessing in order to improve and accelerate subsequent processing steps like semantic video segmentation or video-based object tracking. A framework is proposed to segment video streams into temporally consistent superpixels in order to create a representation of the video with far less image primitives than the voxelgrid. The proposed energy-minimization-based approach utilizes a novel hybrid clustering strategy for a multidimensional feature space. Techniques are presented to ensure the consistency of the superpixel flow with the image movement while considering visual occlusion and disocclusion effects. The effectiveness of the proposed method is shown by a comparison to state-of-theart spatio-temporal oversegmentation algorithms using established benchmark metrics. Additionally, its effectiveness is further demonstrated by showing its application on the real-world scenario of interactive video segmentation. Kurzfassung Eine ...