Advanced algorithms for manipulating 2D objects on touch screens
Permanent address of the item is
Summary ViewShow full item record
Edistyneitä algoritmeja kaksiulotteisten kappaleiden muunteluun kosketusnäytöllä
Multi-touch gestures are widely used on touch screens to scale, rotate, or otherwise transform geometric objects such as maps and images. Typical gesture recognition implementations handle one or two touch points successfully but work in various unnatural and error-prone ways with additional fingers or users which are increasingly common due to growing screen sizes. We claim this deficiency to originate from the lack of software developer friendly material for estimating transformations from unlimited and changing number of touch points. We attempt to correct that by deriving 7 reliable and fast algorithms to estimate an optimal translation, rotation, scaling, or one of their combinations from any number of touch points. Mathematically this reduces to solving an optimal nonreflective similarity transformation matrix through the least squares method for each combination. We present the algorithms in implementation-ready source code, show them to be computationally efficient, and implement them in a production-ready software package called Nudged.