Hardware/Software CoDesign of an Acceleration SoC for Interactive Digital Painting

Abstract

Novel digital paint systems use brush-like or real brush input for computer assisted graphics- and art-creation. This paper presents the hardware-software implementation of the real-time video processing algorithms for brush position, shape and bristle texture detection. The system has been implemented using a 5 megapixel camera and an FPGA system. The system consists of a dynamically reconfigurable image processing streaming architecture. It is heavily pipelined to keep the propagation delays within the strict timing requirements in order to enable the processing at camera video rate. Video processing blocks are realized in a modular way and interconnected by a dedicated programmable switch fabric. A 32 bit RISC processor controls the settings and parameters of the image processing, the camera settings as well as the communication with the host PC. Segmented brush footprints are forwarded by an Ethernet link to the paint rendering host PC. The system can be used for real-time virtual brush based painting.