Design of algorithm for detection of hidden objects from Tera hertz images
IOSR Journal of Computer Engineering · 2013
An algorithm for detection of hidden objects from tera hertz images is presented.Presently, terahertz imaging employs object's radiometric temperatures in human to acquire images of concealed objects.But, it presents problem in temperature sensitive areas like oil and coal mines, factories etc.The aim of the paper is to detect and extract hidden objects underneath person's clothing.Here, a three stage approach is presented: In the first stage, edge based segmentation is applied after smoothing the image using bilateral filter.In the next stage, transform invariant shape descriptors, Gabor and gray level co-occurrence (GLCM) texture features of interested object regions are computed.Finally, a Euclidean distance criterion is used for classification.To appraise the technique, detection error and detection rate are calculated.Test results are compared with ground truth data obtained from the original image.Experiment results are found to be promising with 1.04% detection error and 91.9% as detection rate.Potential applications in security include detection of weapons and explosive in public places like airports, stations etc.