Automatic Pedestrians Detection System Based of Features Level Extraction
Reda Shbib, Ali S. Rachini · 2019
The estimation of Crowd density and counting have become mow an interesting area for most of scientists, for instance assessing the social consequence and influence among minor groups of persons inside a crowd. Still, current investigational crowd investigates achieved by workers are costing in term of time. Usually, human is involved to attain this job, yet, increasingly, visual observation is now a vital requirement, it is a tough job to monitor and asses all documented video due to the massive amount of cameras that have been used. Now, image-processing zone has become a concern of all educational and study to improve or investigate new protocol or technique in order to automated and facilitate this task of counting and monitoring with less human intervention. In this paper, novel techniques in that areas have been implemented using different datasets to asses and validate our results. This paper has presented a novel technique which is based on replacing of some global based features with low based level features, which are precise to persons and groups reside in a specific crowd. Therefore, the entire sum of walkers is the total sum of all groups summed together, which build the entire crowd.