Zhiyi Yu and colleagues at China's Sun Yat-sen University have developed a new hand gesture recognition algorithm for easy application in consumer-level devices.
The algorithm can adapt to different hand types by classifying them as slim, normal, or broad, based on measurements of palm width, palm length, and finger length.
The algorithm compares a gesture with stored samples of the same hand type, through which "we can improve the overall recognition rate with almost negligible resource consumption," said Yu.
The program also can pre-recognize likely input gestures and choose the final gesture using feature extraction, which Yu said shortens recognition speed without losing accuracy.
The method was able to recognize hand gestures in real time with more than 93% accuracy, even when input images were rotated, translated, or scaled.
From SPIE Newsroom
View Full Article
Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA
No entries found