FPGA及手勢辨識為基礎之輪型助行器引導

由於台灣目前正逐步邁向高齡化社會，針對老年人的居家照護設備需求也正逐年成長。本研究在實際審視老人復健所需之助行器後，提出一款以晶片設計為主之手勢指揮辨識系統。此系統主要是以場域可編輯邏輯閘陣列(Field Programmable Gate Array, FPGA)為晶片開發之平台。所設計之功能包含移動目標偵測、色彩鑑別以及模糊理論應用。藉由偵測手勢揮動之移動軌跡來指揮助行器之行駛方向，以確保老人復健安全。本系統與一般以圖形辨識(Pattern Recognition)之手勢辨識相比，本研究是利用移動偵測(Moving Detection)來完成手勢之追蹤。其主要優點在於移動偵測僅需運用簡單之處理程序，而且偵測率較高。此外，本研究特色還包含了快速單頁像素比較演算法(One-Page-Comparison , OPC)，使得設計大幅減少記憶體空間的需求量，並且能夠在影像掃描的過程中即可完成處理。而手勢辨識的結果可以運用模糊理論(Fuzzy Theorem)來加以推論實際指揮揮動幅度之大小，以當作助行器行駛策略之參考。最後，經由實際的測試以及實驗結果顯示，兩輪式助行器參考速度差在透過模糊控制器運算輸出之後，相較於實際的理想值，其最大誤差率不超過15%，且平均誤差率均低於10%。此結果不但完全符合系統設計之預期結果，本系統之設計概念在未來亦可運用在其他移動式平台之半自動化行駛控制。

關鍵字

快速單頁像素比較演算法；手勢辨識；場域可編輯邏輯閘陣列；移動偵測；模糊理論

並列摘要

Due to the deterioration of aging society in Taiwan, the demands for homecare facilities have been increased. After surveying the capability of walking aids devices, in this research, it especially develops a gesture directing recognition system for elderly people. Here proposed system is implemented on a Field Programmable Gate Array (FPGA) chip for prototyping, which involves functions of moving object detection, color discrimination, and Fuzzy logic. Integrating all designs together provide an efficient gesture directing system for elders’ rehabilitation on expected direction. Comparing to traditional gesture recognition by pattern recognition, proposed chip-based system is using moving detection for gesture tracking. The advantage of such design is for lower computing load and higher detecting rate. Moreover, by adopting the mechanism of One-Page-Comparison (OPC) algorithm, gesture recognition can be performed with minimal memory space during image scan. The result of gesture directing recognition can be classified by Fuzzy logic for real-life directing operation, which is fed to an autonomous walking aids platform for steering strategy. Finally, according to experiment results of the Fuzzy output to wheeled differential robot, the maximum error has been efficiently controlled within 15%, comparing to theoretical estimation, and 10% is for average error. This result is sufficient to our expectation for real-world operation. It can be seen proposed system is viable for semi-automatic mobile platform control in the future.

並列關鍵字

One-Page-Comparison Algorithm ； Fuzzy Theorem ； moving detection ； Field Programmable Gate Array ； gesture recognition

參考文獻

Wang, X., Xia, M., Cai, H., Gao, Y., and Cattani, C., “Hidden-Markov-Models-Based Dynamic Hand Gesture Recognition,” Mathematical Problems in Engineering, 2012.

Adams, J., Parulski, K., and Spaulding, K., “Color Processing in Digital Cameras,” IEEE Micro, Vol. 8, issue 6, pp. 20-30, 2002.

Capp´e, O, Moulines, E., and Ryd´en, T., Inference in Hidden Markov Models, New York: Springer, 2005.

Chen, C-W., and Fuh, C-S., “Lens Shading Correction for Dirt Detection,” Pattern Recognition, Machine Intelligence and Biometrics,Berlin Heidelberg, Germany, pp. 171-195, 2011.

Chen, T-H., Yin, Y-H., Huang, S-F., and Ye, Y-T., “The Smoke Detection for Early Fire-Alarming System Base on Video Processing,” Proc. Int. Conf. on Intelligent Information Hiding and Multimedia Signal Processing(IIH-MSP, 2006), Pasadena, USA, pp. 427-430, 2006.

國際替代計量

FPGA及手勢辨識為基礎之輪型助行器引導

未授權

主題瀏覽