The current state-of-the-art hand gesture recognition methodologies heavily rely in the use of machine learning. However there are scenarios that machine learning cannot be applied successfully, for example in situations where data is scarce. This is the case when one-to-one matching is required between a query and a dataset of hand gestures where each gesture represents a unique class. In situations where learning algorithms cannot be trained, classic computer vision techniques such as feature extraction can be used to identify similarities between objects. Shape is one of the most important features that can be extracted from images, however the most accurate shape matching algorithms tend to be computationally inefficient for real-time applications. In this work we present a novel shape matching methodology for real-time hand gesture recognition. Extensive experiments were carried out comparing our method with other shape matching methods with respect to accuracy and computational complexity using our own collected hand gesture dataset and a modified version of the MPEG-7 dataset.%that is widely used for comparing 2D shape matching algorithms. Our method outperforms the other methods and provides a good combination of accuracy and computational efficiency for real-time applications.
翻译:目前最先进的手势识别方法在很大程度上依赖于机器学习的运用。 但是,有些假设是机器学习无法成功应用的, 比如在数据稀少的情况下。 当每个手势的查询和数据集之间需要一对一匹配时, 每个手势代表一个独特的类别。 在无法培训学习算法的情况下, 经典的计算机视觉技术, 如特征提取等, 可以用来辨别对象之间的相似性。 形状是可以从图像中提取的最重要特征之一, 但是最精确的形状匹配算法往往在计算上对实时应用无效。 在这项工作中, 我们提出了一个新的形状匹配方法, 用于实时手势识别。 进行了广泛的实验, 利用我们自己收集的手势数据集和修改版的 MPEG-7 数据集, 将我们的方法与其他形状匹配方法进行比较, 准确性和计算复杂性。% 用于比较 2D 形状匹配算法时, 我们的方法优于其他方法, 并为实时应用提供了精准和计算效率的良好组合 。