- Jazem Mutared Alanazi
Assistant Professor, Computer Science Department, Community College, King Saud University
Effective Machine-Learning Based Traffic Surveillance Moving Vehicle Detection
The traffic surveillance is an integral part of the traffic monitoring. Automatic vehicle recognition is mostly used in efficient traffic management systems. Basic indicators of a successful traffic monitoring system are its resilience and dependability. In this study, we replace the pixel-by-pixel approach with a block-level technique, which drastically decreases the computing time and complexity by using a transform domain which naturally shortens the process's runtime. This is because discrete cosine transforms (DCT) makes it simple to extract the variations in intensities that are characteristic of many ecosystems. In most cases, the low-frequency component is where the most important information is stored inside the DCT blocks since it is less susceptible to noise. In this acquiring the edge shape feature of the detected item alongside the texture feature that provides the obvious moving top and bottom, the discriminative robust local ternary pattern (DRLTP) edge extraction is suggested which indistinct limits that improve detection efficiency. By summing the values of robust local ternary pattern LTP (RLTP) and differential LTP, we get the DRLTP, which is used for edge extraction (DLTP). RLTP is the greatest absolute value of LTP and its complement, whereas DLTP is the greatest absolute value of the difference between LTP and its complement.