A parallel VLSI architecture of contactless HCI system

This paper presents a field programmable gate array (FPGA) based 2-dimentional hand gesture recognition system using Haar feature Adaboost detector, skin color segmentation, and frame subtraction background modeling. The system obtains detection results and tracking results simultaneously. In addition, the interaction instructions are interpreted by a judgment strategy. The proposed system can complete the processing of 640*480 resolution video at 60 frames per second. The design has been synthesized and mapped for the Altera Stratix IV board. The proposed architecture compensates for the low detection speed of software implementation systems for the hardware implementation utilizes the parallelism inherited in the algorithms.