WinVMDS integrated with neural recognition windows-based video motion detector system integrated with neural recognition

• Main Authors: Ang, Rosalinda C., Cruz, Maria Corina L., Tan, Lilybeth U., Yap, Maria Paula R.
• Format: text
• Language: English
• Published: Animo Repository 1994
• Online Access: https://animorepository.dlsu.edu.ph/etd_bachelors/16595
• Institution: De La Salle University

WinVMDS Integrated with Neural Recognition is a neural network-based recognition system in which the main object is to create a microprocessor-based system for building or home security. The system is divided into several modules. The Ultrasonic Module is responsible for detection thereby sending a signal to trigger the system. When and if triggered, the video camera captures the image. The image captured and a pre-captured image, that is an image of the plain background, passes through the Feature Extraction Module to extract the image that triggered the system. Hereafter, the image extracted is prepared for input to the neural network by the Image Enhancement module. The resulting image is stored in a file of 1's and 0's representing the black and white pixels respectively. The file, before being entered into the main recognition module is segmented into 100 divisions, a procedure adapted to compress and convert the image into an acceptable form. The Neural Recognition Module then determine if the initially captured image exhibits a form of a human being. If the image triggers a yes in the neural recognition module, the Siren Module is triggered. The system was trained on 50 training sets equally divided to the different orientations of the human body and non-human images. Increase in the number of training sets and decrease in acceptable error would lessen the probability of the neural recognition module in bringing about an error.