转自：http://www.altera.com/education/univ/events/innovate_contests/na/unv-innovate-na-2010.html

Altera^® would like to thank all contestants who entered the 2010 Innovate North America contest. There were many interesting and innovative projects. The top five finalists are described below.

University of Saskatchewan
1st Place

Real-Time American Sign Language (ASL) Recognition System
Zhongkai Chen, Xiuxin Yang and Ariq Chowdhury, supervised by professor Hsiang-Yung (Daniel) Teng

The winning project was a real-yime american sign language (ASL) recognition system. The system recognizes ASL as well as functional gestures to play music or take a picture or operate arithmetic calculation. Although different kinds of sign-language recognition systems have been implemented, most of them are based on data-glove. Since the data-glove is expensive (the VPL data-glove costs over $9,000), a new solution using less expensive (only $1) color-glove was devised. When the system recognizes a relevant gesture it converts it to sound using speakers, playing a pre-recorded WAV file and outputs the results onto VGA monitor. In addition, the sign language also provides a different way for the human beings to interact with computers, an innovative sign-language-based music player is implemented in this system: users can control the music play by using different signs, instead of the usual push-button operation.
Download the project: UniversityOfSaskatchewan.zip

University of South Carolina
2nd Place

Acceleration of Frequent Itemsets Mining
Yan Zhang and Zheming Jin, supervised by professor Jason D. Bakos

The team created an FPGA system to accelerate the mining of frequent itemsets based on the Apriori algorithm. The system includes both a soft-core processor and logic functions implemented in the FPGA fabric. Using ImpulseC's C-to-FPGA compiler, the team partitioned the software and hardware, defined and tested the interfaces prior to implementing in hardware. Software and hardware interaction was successfully simulated even before hardware was generated. Profiling was used to identify which parts of the program were most expensive in terms of execution time; once the bottleneck was found, the system was optimized to solve the problem.
Download the project: UniversityofSouthCarolina.zip

Purdue University
3rd Place (tied)

Power-aware Image Processing System
Georgios Karakonstantis, Vaibhav Gupta and Himanshu Markandeya, supervised by professors Kaushik Roy and Anand Raghunathan

The team created a power-aware image processing system that uses the novel concept of adaptive quality modulation based on significance driven computation. On the basis that in DSP applications not all computations are equally important in determining the output quality, they identified crucial operations that were imperative for providing a good output quality. The system discards computations that contribute the least to the output image quality, thus reducing the number of cycles required which in turn reduces the energy consumed by the overall system. The team's scheme departs from conventional approaches which mainly modify the resolution of the processed images in order to increase performance or reduce power. Users can capture the image with high quality and later decide on power or quality modes.
Download the project: PurdueUniveristy.zip

McMaster University
3rd Place (tied)

3D Camera Anaglyph System
Charles Wah, Khaled Chebaro, Lucas Dobrowolski and Yaser Jafar, supervised by professors Mohamed Bakr and Xun Li

The team created a 3D anaglyph camera system that involves interfacing two CMOS image sensors with the DE2 Altera board, storing the data on SDRAM, and then applying image processing before being output through VGA to the monitor. The group decided to utilize anaglyph technology, as it is currently the most practical solution in terms of cost and application as well as have the ability to view the 3D experience offline unlike other technologies. The team decided to pursue this project as the 3D technology is at its early stage of growth in all market segments and there is future potential in terms of applications. The product would target users in the desktop and laptop market as a webcam for use in popular instant messaging programs. It can also be used in digital cameras, camcorders and cell phone cameras.
Download the project: McMasterUniversity.zip

University of Ottawa
Honouable Mention

Bidirectional Human Computer Interface
Lukash Monczak, Alexey Borisenko, Ikvir Singh and Ovidiu Draghici, supervised by PhD candidate Daniel Shapiro

The team created a bidirectional human computer interface that interacts with the human body using a tongue display unit (TDU). Human Computer Interface (HCI) devices are being used by hospitals and clinics to alleviate the suffering of patients that have lost vision, balance, or mobility. The team's HCI is composed of low cost surface mounted circuit board components, instead of a high-cost integrated circuit. One example of a commercial HCI is BrainPort, which allows patients with vestibular balance disorders to regain their sense of balance through neuroplasticity. The team's project reduces the cost and effort of training by using a the Pong computer game and their TDU to make the learning process low cost and fun. The automation reduces the cost of the system by relying less on human trainers, and instead emphasizing game play. The patient feels the game using the team's TDU, and plays the game by using the Altera DE1 board push buttons. An audience can watch the game being played via the DE1 VGA interface, while the player can only sense the game on his/her tongue.
Download the project: UniversityOfOttawa.zip