Research On The Algorithm Of Multi-touch Gesture Recognition

Multi-touch technology is the next generation of interactive technology, itsupports one or more fingers to operate at the same time. It is proved that operate byhands via the touch screen has a high degree of freedom interact as well. With thedevelopment of3G wireless communication technology, the mobile Internet era hascome, the prevalence of MID (Mobile Internet Devices) equipment has become aninevitable trend, while the touchscreen, multi-touch screenespecially, will become amainstream input device.Multi-touch technology is more mature in terms of hardware, but forthe way ofmulti-touchmanipulation is relatively immatureat home and abroad, especially lackthe extension of multi-touch operating gestures, which is a serious restrictions on thefurther application of multi-point operation, therefore, the definition of multi-touchgestures, as well as the optimization and improvement of its recognition algorithmnow has an important practical significance.Therefore, this article will focus on theproblems above-mentioned on multi-touchgesture recognition today, and made some research: First, through investigation andanalysis of user habits,according to human-computer interaction theory, then proposedthe most optimized user experiencegesture trajectory and the corresponding gesturecommand, and then propose comprehensive proposed an efficient, simple multi-touchgesture design.At the same time, base on thecharacteristics of multi-touch gesturemanipulation, we will propose a correct and effective data extraction plan, so thatprepare for the next recognition algorithm model.Secondly, according to the design of multi-touch gestures, propose themulti-gesture recognition model. According to the characteristics of multi-pointgestures, through the combination of BPneural network and multi-point gestures andmathematical modeling and process analysis, put forward a comprehensivemulti-point gestures of neurons and neural network model, gives the process ofmulti-point gestures of the training algorithm, and the gesture recognition process.Atthe end,implement the entire algorithm model and a multi-touch gesture recognitionalgorithm prototype program. Finally, we will analysis the flaws and shortcomings of the BP neural network,research to find out the reason for the existence of defects.Then summarize andpropose effective solutions, combined with the prototype program, set up and toimplement an experimental platform.Use the experimental statistics and validationalgorithm experimental data to prove the effectiveness of the algorithm and theaccuracy of identification. So that this algorithm will make the advantageofmulti-touch operation further play an important role.