Real-time recognition of hand-drawn patterns using an innovative web-based approach

An innovative system is proposed for using mobile devices for remote recognition. When the user draws a pattern in the local browser, the pattern is transmitted to a remote server that immediately performs the recognition task and database search. The recognition system uses bidirectional associative memory (BAM) with artificial neural network technology. This study is based on the example of buying a dog house on the network. When the mouse or finger is used to draw an outline of the doghouse in the browser, the proposed system immediately recognizes the drawing and displays doghouses on the web page and allows users to pick them. First, this study classified commercial doghouses and set the corresponding graphics stored in the system. The system are also the graphics as training sample patterns. Current networks almost always require users to enter key words to search for relevant information. The innovative approach proposed in this study enables the use of mobile devices for recognition of hand-drawn graphics. Finally, the result of experiments performed to measure pattern recognition accuracy are discussed.