A Dynamic Gesture Recognition System based on CIPBR Algorithm

Abstract

Dynamic gesture recognition has been studied actually for it big application in several areas such as virtual reality, games and sign language. But some problems have to be solved in computer applications, such as response time and classification rate, which directly affect the real-time usage. This paper proposes a novel algorithm called Convex Invariant Position Based on Ransac which improved the good results in dynamic gesture recognition problem. The proposed method is combined with a adapted PSO variation to reduce features and a HMM and three DTW variations as classifiers.

FSSGR: Feature Selection System to Dynamic Gesture Recognition

“…In Table 2 is presented the proposed system results in comparison with previous systems based on LCS [13] and SURF [14], as well the extensions based on the Convexity Approach: CLCS and CSURF [7]. Finally, it is presented the results achieved by Santos et al [6] with CIPBR technique. All works used the same dataset and HMM as classifier.…”

“…. , p n } from Convex hull points, which is used to generate two signature sets, according to the CIPBR approach [6]. The first signature set is composed by distances (D ωQ ) calculated by the following way.…”

“…We select the perfect value by defining this size of our feature vector as the function to be optimized by a Swarm technique. Additionally, this work proposes a feature selection system applied in gesture recognition based on the CIPBR algorithm [6] in combination with a Binary Particle Swarm Optimization [5] (BPSO) and a selector algorithm proposed by Barros et al [7]. Our proposed system, named Feature Selection System to Dynamic Gesture Recognition (FSSGR), aims to generate small feature vectors improving the classification rate and speeding up the system response.…”

HAGR-D: A Novel Approach for Gesture Recognition with Depth Maps

“…In this paper, we propose a novel approach for dynamic gesture recognition with depth maps, called the hybrid approach for gesture recognition with depth maps (HAGR-D). HAGR-D uses a version of CIPBR (convex invariant position based on RANSAC) algorithm [ 33 ] for feature extraction, a combination of the binary particle swarm optimization [ 34 ] and a selector algorithm to make the feature selection and a hybridization between DTW and HMM classifiers for recognition. DTW is used to find the most probable gestures, while HMM refines DTW output.…”