An integrated mobile system for gang graffiti image acquisition and recognition
In this thesis we describe a system that allows first responders to collect and browse images of gang graffiti for the purpose of identifying and tracking gang activity. This system is implemented both as an application for Android hand-held devices and as a web-based interface. ^ Our system includes the acquisition of gang graffiti images and associated metadata. The data obtained from the device is transferred to a database. We also developed methods to provide basic color feature interpretation of gang graffiti images. We describe a color recognition method for hand-held devices based on touchscreen tracing and a color image segmentation method based on Gaussian thresholding. The segmentation creates a probability map that can be used for shape recognition and analysis. A user can browse the image database by graffiti color, location, date and time, with the use of interactive map projections. Additionally, a user can manually add information to the database, such as gang names, symbols, or general comments.^
Edward J. Delp, Purdue University, Mireille Boutin, Purdue University.
Engineering, Electronics and Electrical