An edge detector based integrated database framework for security applications

Abstract

In this paper, an integrated framework comprising of computer vision algorithms, Database system and Batch processing techniques has been developed to facilitate effective automatic threat recognition and detection for security applications. The proposed approach is used for automatic threat detection. The novel features of this structure include utilizing the Human Visual System model for segmentation, and a new ratio based edge detection algorithm that includes a new adaptive hysteresis thresholding method. The feature vectors of the baseline images are generated and stored in a relational database system using a batch window. The batch window is a special process where image processing tasks with similar needs are grouped together and effectively processed to save computing and memory requirements. The feature vectors of the segmented objects are generated using the CED method and are classified using a support vector machine (SVM) based classifier to identify threat objects. The experimental results demonstrate the presented framework efficiency in reducing the classification time and provide accurate detection.