Hand Shape Recognition Research Articles

Hand Shape Recognition in Human Machine Interaction through the Singularity Detection with Wavelet Transform

This paper represents a hand shapes recognition system for the Human Machine Interaction (HMI) with service robot of disable people. This system uses a touchpad to precept the touching of fingers, as well as to provide a background for hand shapes image. Each finger can stay in one of the 4 statuses: stretch- touching on the pad, retracting-touching on the pad, stretch-detaching over the pad and retracting-detaching over the pad. Hand shapes, posed to express HMI instructions, are defined by the status combinations of Index finger, Middle finger, Ring finger and Little finger. Hand shape features, the relative heights of the fingertips, are extracted through the singularity detection with wavelet transform on hand shape contour. The hand shape recognition of this system is based on an optimized Bayesian decision binary tree. The design of 2 types of classifier in the tree and the corresponding error rates of the classifiers are analyzed. Implemented by a DSP processor, a correctness ratio of over 98% is obtained in the identification of 12 hand shapes. Experiments show that this system can provide a flexible, humanized and expendable HMI for service robot, as well as for other applications.

Mobile robot control using hand-shape recognition

This paper presents a vision-based walking robot control system using hand-shape recognition. To recognize hand shapes, the accurate hand boundary needs to be tracked in an image obtained from a moving camera. For this, we use an active contour model-based tracking approach with mean shift, which reduces the dependency of the active contour model on the location of the initial curve. The proposed system consists of four modules: a hand detector, a hand tracker, a hand-shape recognizer and a robot controller. The hand detector detects a skin-colour region, which has a specific shape, as a hand in an image. Then, hand tracking is performed using an active contour model with mean shift. Thereafter, hand-shape recognition is performed using Hu moments. To assess the validity of the proposed system, we tested the proposed system on a walking robot, KHR-1. The experimental results show the effectiveness of the proposed system.

Signing Exact English (SEE): Modeling and recognition

We present effective and robust algorithms to recognize isolated signs in Signing Exact English (SEE). The sign-level recognition scheme comprises classifiers for handshape, hand movement and hand location. The SEE gesture data are acquired using CyberGlove ® and magnetic trackers. A linear decision tree with Fisher's linear discriminant (FLD) is used to classify 27 SEE handshapes. Hand movement trajectory is classified using vector quantization principal component analysis (VQPCA). Both periodic and non-periodic SEE sign gestures are recognized from isolated 3-D hand trajectories. Experiments yielded average handshape recognition accuracy of 96.1% on “unseen” signers. The average trajectory recognition rate with VQPCA for non-periodic and periodic gestures was 97.3% and 97.0%, respectively. These classifiers were combined with a hand location classifier for sign-level recognition, yielding an accuracy of 86.8% on a 28 sign SEE vocabulary.

Associative learning method in a hypercolumn model

We propose an associatively learnable hypercolumn model (AHCM). A hyper-column model is a self-organized, competitive, and hierarchical multilayer neural network. It is derived from the neocognitron by replacing each S cell and C cell with a two-layer hierarchical self-organizing map. The HCM can recognize images with variant object size, position, orientation and spatial resolution. However, feature maps may integrate some features extracted in the lower layer even if the features are extracted from input data which belong to different categories. The learning algorithm of the HCM causes this problem because it is an unsupervised learning used by a self-organizing map. An associative learning method is therefore introduced, which is derived from the HCM by appending associative signals and associative weights to traditional input data and connection weights, respectively. The AHCM was applied to hand-shape recognition. We found that the AHCM could generate an appropriate feature map and higher recognition accuracy compared with the HCM.

Processing the Shadow of a Hand Image

In the field of the personal authentication system, types of hands are popularly used to certificate the identity of a person. There is an important part in the hand-shape recognition system. That is how to take the photographs of hands. Whether you use a camera or a scanner to take a photograph, there exist some shadow parts around a hand image. As for this part, to judge whether the shadow is true or it is just the hand plays an important role to improve the recognition rate. Actually if we can judge the range of the shadow of the hand image clearly we can raise the probability of the certification. In this paper we focus on the shadow part of hand images. We propose some algorithm to convert an original image into a better image to handle the hand-shape. Firstly, we make the convert the input color image into a grayscale image, then we extract the outline by using Fourier transform, and the true outline of the hand will be determined. Now the recognition methods using biometrics are spread in the world, we believe our project of processing the shadow part will be useful to the recognition system.

Model-guided deformable hand shape recognition without positioning aids

This work addresses the problem of deformable hand shape recognition in biometric systems without any positioning aids. We separate and recognize multiple rigid fingers under Euclidean transformations. An elliptical model is introduced to represent fingers and accelerate the search of optimal alignments of fingers. Unlike other methods, the similarity is measured during alignment search based on finger width measurements defined at nodes by controllable intervals to achieve balanceable recognition accuracy and computational cost. Technically, our method bridges the traditional handcrafted-feature methods and the shape-distance methods. We have tested it using our 108-person-540-sample database with significantly increased positive recognition accuracy.

Hand Shape Recognition using Higher Order Local Autocorrelation Features in Log Polar Coordinate Space

The friendly communication can be more promoted between the human and computer if the function of gesture recognition is implemented to the computer system as the input interface along with the keyboards and mice. We propose a mouse-like function for estimating hand shape from input images with a monocular camera, with which a computer user feels no restraint or awkwardness. Our system involves conversion of sequential images from Cartesian coordinates to log-polar coordinates. Temporal and spatial subtractions and color information are used to extract the hand region. The origin of log-polar coordinates is chosen as the center of the acquired image, but once the hand has been extracted, the estimated centroid position of the hand region in the next frame, obtained from the current hand position and speed, is used as the origin to convert. Recognition of the hand shape is carried out by multiple regression analysis using higher order local autocorrelation features of log-polar coordinate space. Mouse-like functions are realized according to the hand shape and motion trajectory. Compared to conventional Cartesian coordinates, conversion to log-polar coordinates enables us to reduce image date and computation time, remove the variability by the scaling, and improve antinoise characteristics.

Hand Shape Recognition Using the Bending Element of Fingers in Sign Language.

In this study, at first, it classifies hand shape of sign language in order to more efficiently recognize the hand shape, and extracting the element of the hand shape, it is confirmed that bending element of each finger is the most important for the sign language. Next, the recognition method which classifies the each finger bending condition into 5 states is proposed, as to give the detailed erratum judgement of what kind of condition the each finger may be for the learner in hand shape of sign language, is possible. In addition, the recognition method by error reverse propagation method of neural network is proposed as hand shape measuring device for the hand shape recognition in order to recognize the hand shape in which there is individual difference using data glove which can measure the bending angle of the joint of the finger. Then, the effectiveness of the technique that carries out and proposes the commercial hand shape recognition experiment is confirmed.

Hand shape recognition

Automated hand shape recognition has been studied over the past decade and some commercial systems have been developed. Despite these advances, there is not much open public literature discussing the hand shape verification research. This study proposes a method by using quadtree techniques, which are able to recognize the hand shape image within an extremely short time. The geometrical shape of a hand is a biometric characteristic of human beings, although it is different even for a twin sibling. This study uses a parallel grating to project onto the backside of a hand. The parallel grating will be distorted by the curvature shape of the hand and processed by image processing techniques for recognition. This study also presents our recognition results of 100 students captured over a period of time.

Relation between object recognition and formation of hand shape: A computational approach to human grasping movements

We investigate the relation between object recognition and formation of hand shape in human grasping movements through psychophysical experiments. There were few differences between the experimental data when the subjects formed the hand shape without touching objects and when the subjects actually grasped the objects. Both data included information related to the size and shape of the objects. The experimental results suggest that humans are able to compute the hand shape needed to grasp the objects from their visual properties alone before actual grasping. Moreover, developing a five-layered neural network model, we confirmed that the hand shape can be computed from only visual information in the feed-forward manner and that the property information of the objects is extracted when visual-motor transformation is performed. © 2000 Scripta Technica, Syst Comp Jpn, 31(12): 11–22, 2000

A Comparison of Imitation Strategies in Observational Learning of Action Patterns

The effects of different arrangements of demonstration and imitation of modeled actions on the learning of the 26 handshapes of the American manual alphabet were investigated. A concurrent group (N =16), which imitated handshapes concurrently with their demonstration, was compared with a delayed group (N = 16), which delayed imitation until 3 handshapes had been displayed, and with a combination group (N = 16), which practiced under a combination of concurrent conditions early in acquisition and delayed conditions later in acquisition. Following acquisition, learning was assessed by means of immediate and long-term recall and recognition tests. The delayed group was superior to the concurrent group in long-term serial recall and in immediate and long-term recognition of 3-letter sequences (in nonserial order); the performance of the combination group was between those of the delayed and concurrent groups. Therefore, delaying imitation in acquisition required subjects to expend more cognitive effort to retain and produce handshapes when requested than did concurrent imitation. This was beneficial to development of task knowledge that could be relied on for postacquisition recall and recognition of handshapes.

Training with Images: Real and Representational

A total of 60 subjects (24 males and 36 females) participated in a study to determine if a caricature's accentuation of critical cues results in improved recognition of handshapes used in the American Sign Language manual alphabet. Each subject was randomly assigned to one of two training groups. One group trained with photographs of the handshapes and the other group used caricatures. Once mastery of the alphabet was demonstrated, their ability to recognize the handshapes shown in four different modes (positive and negative photographs, and positive and negative caricatures) was tested. In the unrestricted condition, the duration of exposure was not artificially constrained. In the restricted testing condition, the handshapes were displayed at 320 msec, 500msec and 700 msec. Both speed and accuracy were equally emphasized in the training and in the testing. The findings did not support the superfidelity hypothesis of caricatures. In the unrestricted condition, those trained with photographs responded significantly faster, regardless of mode, than those trained with caricatures. As predicted in the most restrictive display time (320msec), mean response time was significantly faster with caricatures. This study has direct implications regarding the media used to train American Sign Language. The findings support the use of photographs to depict and to train novices in the ASL handshapes. Further research is needed to determine if these findings hold true as the complexity of the handshape increases.

手形状の認識に基づき実時間で反応する人物表情の動画像合成

手形状の認識に基づき実時間で反応する人物表情の動画像合成