Human Gait Recognition (HGR) is a biometric approach, widely used for security purposes from the past few decades. In HGR, the change in an individual walk along with wearing clothes and carrying bag are major covariant controls which impact the performance of a system. Moreover, recognition under various view angles is another key challenge in HGR. In this work, a novel fully automated method is proposed for HGR under various view angles using deep learning. Four primary steps are involved such as: preprocessing of original video frames, exploiting pre-trained Densenet-201 CNN model for features extraction, reduction of additional features from extracted vector based on a hybrid selection method, and finally recognition using supervised learning methods. The extraction of CNN features is a key step in which our target is to extract the most active features. To achieve this goal, we fuse the features of both second last and third last layers in a parallel process. At a later stage, best features are selected by the Firefly algorithm and Skewness based approach. These selected features are serially combined and fed to One against All Multi Support Vector Machine (OAMSVM) for final recognition. Three different angles 180, 360 and 540 of the CASIA B dataset are selected for the evaluation process and accuracy of 94.3%, 93.8% and 94.7% is achieved respectively. Results show significant improvement in accuracy and recall rate as compared to the existing state-of-the-art techniques.

Prosperous Human Gait Recognition: an end-to-end system based on pre-trained CNN features selection

Falls among the elderly comprise a major health problem. Daily activity monitoring and fall detection using wearable sensors provide an important healthcare system for elderly or frail individuals. We investigated the classification accuracy of daily activity and fall data based on surface electromyography (sEMG) and plantar pressure signals. sEMG and plantar pressure signals were collected, and their features were extracted. Suitable features were selected and combined for posture transition, gait, and fall using the Fisher class separability index. A feature-level fusion method, named as the global canonical correlation analysis of weighting genetic algorithm, was proposed to reduce dimensions. For the problem in which the number of daily activities is considerably more than the number of fall activities, Weighted Kernel Fisher Linear Discriminant Analysis (WKFDA) was proposed to classify gait and fall. Double Parameter Kernel Optimization based on Extreme Learning Machine (DPK-OMELM) was used to classify activities. Results showed that the classification accuracy of the posture transition is 100%, and the accuracy of gait and fall classified using WKFDA can reach 98%. For all types of posture transition, gait, and fall, sensitivity, specificity, and accuracy are over 96%.

/pdf/daily-activity-monitoring-and-fall-detection-based-on-2bvg8w9k5t.pdf

Daily Activity Monitoring and Fall Detection Based on Surface Electromyography and Plantar Pressure

A gait event is a crucial step towards the effective assessment and rehabilitation of motor dysfunctions. However, for the data acquisition of a three-dimensional motion capture (3D Mo-Cap) system, the high cost of setups, such as the high standard laboratory environment, limits widespread clinical application. Inertial sensors are increasingly being used to recognize and classify physical activities in a variety of applications. Inertial sensors are now sufficiently small in size and light in weight to be part of a body sensor network for the collection of human gait data. The acceleration signal has found important applications in human gait recognition. In this paper, using the experimental data from the heel and toe, first the wavelet method was used to remove noise from the acceleration signal, then, based on the threshold of comprehensive change rate of the acceleration signal, the signal was primarily segmented. Subsequently, the vertical acceleration signals, from heel and toe, were integrated twice, to compute their respective vertical displacement. Four gait events were determined in the segmented signal, based on the characteristics of the vertical displacement of heel and toe. The results indicated that the gait events were consistent with the synchronous record of the motion capture system. The method has achieved gait event subdivision, while it has also ensured the accuracy of the defined gait events. The work acts as a valuable reference, to further study gait recognition.

https://www.mdpi.com/1424-8220/19/24/5499/pdf

A Determination Method for Gait Event Based on Acceleration Sensors.

Exoskeleton robot is a typical application to assist the motion of lower limbs. To make the lower extremity exoskeleton more flexible, it is necessary to identify various motion intentions of the lower limbs of the human body. Although more sEMG sensors can be used to identify more lower limb motion intention, with the increase in the number of sensors, more and more data need to be processed. In the process of human motion, the collected sEMG signal is easy to be interfered with noise. To improve the practicality of the lower extremity exoskeleton robot, this paper proposed a wavelet packet transform- (WPT-) based sliding window difference average filtering feature extract algorithm and the unscented Kalman neural network (UKFNN) recognition algorithm. We established an sEMG energy feature model, using a sliding window difference average filtering method to suppress noise interference and extracted stable feature values and using UKF filtering to optimize the neural network weights to improve the adaptability and accuracy of the recognition model. In this paper, we collected the sEMG signals of three muscles to identify six lower limb motion intentions. The average accuracy of 94.83% is proposed in this paper. Experiments show that the algorithm improves the accuracy and anti-interference of motion intention recognition of lower limb sEMG signals. The algorithm is superior to the backpropagation neural network (BPNN) recognition algorithm in the lower limb motion intention recognition and proves the effectiveness, novelty, and reliability of the method in this paper.

/pdf/lower-limb-motion-recognition-method-based-on-improved-yuftywexir.pdf

Lower Limb Motion Recognition Method Based on Improved Wavelet Packet Transform and Unscented Kalman Neural Network

Plantar pressure image fusion for comfort fusion in diabetes mellitus using an improved fuzzy hidden Markov model

Surface electromyography (SEMG) is an important tool for analyzing gait movements. Effective segmentation of electromyography (EMG) start/end points is an important step in the analysis of EMG signals. This paper presents a SEMG segmentation algorithm based on signal envelope and integral electromyography. Compared with manual segmentation, the coincidence rate is more than 90%. There is no statistical difference in the characteristic parameters of EMG signals calculated from the start/end points obtained by the segmentation algorithm and the manual segmentation method (P > 0.05). Based on this segmentation algorithm, quantitative analysis and comparison of the force situation of the iliopsoas, musculus gracilis, soleus and tibialis anterior muscles during the complete gait cycle are performed. This study lays the foundation for the application of surface electromyography in the field of rehabilitation analysis and control, such as rehabilitation training and rehabilitation robots.

Research on EMG segmentation algorithm and walking analysis based on signal envelope and integral electrical signal

Accurate measurement of foot shape feature parameters is extremely important in the process of customized shoemaking. A 3D foot’s depth image collected by second-generation Kinect is used to propose a foot shape feature parameter measurement algorithm. Through 3D reconstruction of foot based on improved interactive closest points algorithm, the coordinate transformation, feature point selection, and B-spline curve fitting algorithm, the foot length, foot width, metatarsale girth, and other foot feature parameters were calculated. The 3D foot measurement system using this algorithm is tested, and the results of multiple measurements have a mean variance of less than 0.3 mm. The average error between the algorithm calculation result and the manual measurement result is less than 0.85 mm. The stability and accuracy of the system meet the requirements of custom shoes. It lays a good foundation for the automation and standardization of customized shoemaking.

/pdf/a-3d-foot-shape-feature-parameter-measurement-algorithm-2lj2c7pf7x.pdf

A 3D foot shape feature parameter measurement algorithm based on Kinect2

Research on feature extraction algorithm for plantar pressure image and gait analysis in stroke patients

Plantar pressure image is an important tool for gait analysis. In the gait rehabilitation training of stroke patients, it can provide effective quantitative data reference for doctors to accurately assess the recovery status of patients and formulate targeted treatment programs. In this paper, a gait recognition algorithm for pressure sensing image is proposed. Through DBSCAN (DensityBased Spatial Clustering of Applications with Noise) clustering algorithm and minimum circumscribed rectangular algorithm, footprints can be accurately located in the pressure image for feature extraction. Three different degrees of abnormal gait and healthy gait are trained based on SVM (Support Vector Machine) algorithm. The training results can accurately identify the above abnormal gait, with an average accuracy of 91.4%.

Research on Abnormal Gait Recognition Algorithms for Stroke Patients Based on Array Pressure Sensing System

The acquisition and analysis of plantar pressure can be applied to the postoperative gait rehabilitation training of stroke patients, providing effective quantitative data reference for doctors to evaluate the patient’s condition and formulate targeted treatment plans. In this paper, an insole type plantar pressure sensing system was developed for the needs of long-term continuous monitoring of patients in sports rehabilitation training. The sensing system has a large amount of sensor units (maximum 288) and high spatial resolution (1.64/cm2). Through the optimized circuit structure design, the clear acquisition of pressure sensing images was ensured, and the remote plantar pressure measurement with high sampling rate (100SPS) was realized.

Sixu Zhang

Papers

Research on EMG segmentation algorithm and walking analysis based on signal envelope and integral electrical signal

A 3D foot shape feature parameter measurement algorithm based on Kinect2

Research on feature extraction algorithm for plantar pressure image and gait analysis in stroke patients

Research on Abnormal Gait Recognition Algorithms for Stroke Patients Based on Array Pressure Sensing System

Design and Research of a High Spatial Resolution Insole Plantar Pressure Acquisition System