Positions Of Key Points Research Articles

How to predict the global instantaneous feeling induced by a facial picture?

Picture selection is a time-consuming task for humans and a real challenge for machines, which have to retrieve complex and subjective information from image pixels. An automated system that infers human feelings from digital portraits would be of great help for profile picture selection, photo album creation or photo editing. In this work, two models of facial pictures evaluation are defined. The first one predicts the overall aesthetic quality of a facial image, and the second one answers the question “Among a set of facial pictures of a given person, on which picture does the person look like the most friendly?”. Aesthetic quality is evaluated by the computation of 15 features that encode low-level statistics in different image regions (face, eyes, and mouth). Relevant features are automatically selected by a feature ranking technique, and the outputs of 4 learning algorithms are fused in order to make a robust and accurate prediction of the image quality. Results are compared with recent works and the proposed algorithm obtains the best performance. The same pipeline is considered to evaluate the likability of a facial picture, with the difference that the estimation is based on high-level attributes such as gender, age, and smile. Performance of these attributes is compared with previous techniques that mostly rely on facial keypoint positions, and it is shown that it is possible to obtain likability predictions that are close to human perception. Finally, a combination of both models that selects a likable facial image of good quality for a given person is described.

Least squares image matching: A comparison of the performance of robust estimators

Abstract. Least squares image matching (LSM) has been extensively applied and researched for high matching accuracy. However, it still suffers from some problems. Firstly, it needs the appropriate estimate of initial value. However, in practical applications, initial values may contain some biases from the inaccurate positions of keypoints. Such biases, if high enough, may lead to a divergent solution. If all the matching biases have exactly the same magnitude and direction, then they can be regarded as systematic errors. Secondly, malfunction of an imaging sensor may happen, which generates dead or stuck pixels on the image. This can be referred as outliers statistically. Because least squares estimation is well known for its inability to resist outliers, all these mentioned deviations from the model determined by LSM cause a matching failure. To solve these problems, with simulation data and real data, a series of experiments considering systematic errors and outliers are designed, and a variety of robust estimation methods including RANSACbased method, M estimator, S estimator and MM estimator is applied and compared in LSM. In addition, an evaluation criterion directly related to the ground truth is proposed for performance comparison of these robust estimators. It is found that robust estimators show the robustness for these deviations compared with LSM. Among these the robust estimators, M and MM estimator have the best performances.

Human Body Kinematics and the Kinect Sensor

For a good understanding of human body biomechanics scientist all over the world are using motion capture systems (MoCap). Such a system is mainly composed of one or multiple high performance cameras and a process unit for gathering key information. A low – cost solution for these systems is the Kinect sensor from Microsoft. The Kinect sensor is a depth camera that can be used for assessing full – body movements in terms of joint and/or segment positions and movement geometries. The resulting data can be used in the robotic industry, in clinical solutions, video gaming industries etc. The functionality of the equipment has been highly debated in many studies wherefrom result that the depth camera in question is accurate and reliable in studies such as human biomechanics. The aim of this paper is to explain the logics behind this equipment and its functionality. Therefore we present a new approach in constructing a 3D human skeleton model that can be used for assessing asymmetries by determining the human silhouette in 3D, the position of human body key points and joints, the angles between the track segments and full body kinematics.

Multifractal analysis of wind velocity data

We analyze the wind velocity, v(t), at a location in the Patagonia, Argentina, as a function of time, measured every 10min, uninterruptedly, over a one-year period, by taking the graph (t, v(t)) as a signal, and applying the tools of multifractal analysis. The multifractal spectrum (α, f(α)) of this signal, obtained by the definition of α and f(α), is strangely smooth, for it fulfills all properties of the twice-differentiable thermodynamical formalism, the theoretical algorithm devised to mimic the spectrum-by-definition (with which it may not coincide, since the latter may not even be continuous). We give arguments to conclude that the wind-spectrum is of Weierstrass type. We take a finite sample of Weierstrass functions, Wλ,s(t), at regular intervals Δt, the size of each sample comparable to that of the wind signal. We take these samples as signals, and construct their multifractal spectra by definition, for different parametric values of λ and s. We compare the different Weierstrass spectra with the wind spectrum, and obtain information on the Weierstrass parametric values that agree with the wind spectrum. We interpret the key points of the wind spectrum in meteorological terms. Any change, from one year to the next, in the position of such key points, or in the shape of the spectral curve—once interpreted in the corresponding meteorological terms—could be some local indicator of climate change.

P3‐314: Pictorial education of safe medication for the elderly with cognitive impairment

Background: In this work, we developed a video processing framework, which simultaneously analyzes hands motion of multiple-people to understand the type of action, to measure the reaction time and accuracy of the motion during physical exercise or social activity. The location of the hands, head and center of the body are captured to encode the motion characteristic at each frame from a video sequence. The system can detect multiple people and analyze their hands motion. In the experiments, analysis of various hands motion of two elderly patients is presented. Methods: Recently released Kinect sensor is used to capture depth data and skeleton data along with an RGB camera. The positions of the hands, head, shoulders, elbows and body center are extracted and stored as key points in a 2-dimensional matrix, which is called a location map. Location maps are generated from skeleton views. Later, each location map is compared with a previously defined matrix for each activity and the most similar one is chosen to recognize the type of the activity. The accuracy of the motion is given by the value of the similarity; and the reaction time is given by the time between voice order and realization of the action. Extracting depth data helps us avoid occlusions and easily distinguish people and body parts in the scene. Various activities (Hands Up, Hands Down, Hands on Head, Hand Shaking, Hands Up Arms Straight, Hands Up Arms Bended, Hands Reunited Front, etc.) are encoded in real-time by using the positions of key points in the 2D location map. Results: In the supplementary figure, five different activities (normal position, hands clapping, hands on shoulder, hands up, hand shaking) are investigated for two patients by using the proposed framework. Skeleton views for each frame are extracted and the position of nine key points (head, neck, shoulders, elbows, hands, center) from each skeleton are calculated to compose location maps. Location maps are used to categorize the activity. Conclusions: In this work, a video processing framework to simultaneously capture and categorize hands motion of multiple-people is presented. Experimental results show the potential usage and successful application of the system to analyze five different types of activity. P3-314 PICTORIAL EDUCATION OF SAFE MEDICATION FOR THE ELDERLY WITH COGNITIVE IMPAIRMENT

An evolution strategy in structural optimization problems for plates and shells

For multilayered plated and shell structures the formulation of the optimization problem is strongly dependant on the definition of the design variables. Therefore, the first part of the work is devoted to the definition of design variables and the forms of objective functions. Those design variables define stacking sequences of structures have discrete fiber orientations 0°, ±45°, 90° and a finite number of key points that are required in the evaluation of the curve Γ characterizing an external boundary of the structure or a structural shape understood in the sense of a structural geometry representing a shell/plate mid-surface or thickness distribution of structures. For the curve definition we have adopted one dimensional B-splines. Each curve is formed by an assembly of subsegments passing through certain key points. The positions of key points are randomly generated so that in the generation process it is possible to fulfill the required set of equality or inequality constraints. It is necessary to emphasize that the proposed method is very general and can be applicable to a very broad class of optimization problems. The generality of the approach is confirmed by the proof of the direct equivalence and mapping between discrete fiber orientations and continuous angle ply orientations. The evolution strategy is proposed herein as the optimization algorithm. Similarly as classical ones (e.g. ACO, SS, PS or ISM) it combines all features and advantages of evolution algorithms. It is worth to note that in the evolution strategy the number of children produced in one generation is not limited and it is not necessary to conduct mutation operations as in genetic algorithms. It simplifies significantly the effectiveness of numerical procedures. Then, two numerical examples have been solved to demonstrate the effectiveness of the proposed formulations and the optimization algorithm. They deal with thickness and stacking sequence optimization problems for circular cylindrical shells subjected to various dynamic and static constraints, respectively.

2P2-J12 人の全身姿勢からヒューマノイドの全身姿勢を入力するインタフェースの開発(その1)(ヒューマノイド)

New key poses input system for the Choreonoid that is integrated software to directly generate and edit the whole-body motion of a humanoid robot is developed. Thanks to Kinect^<TM> and OpenNI^<TM>, the system captures human poses without any markers and get position and orientation data of key points of the human body from the pose. The data is converted to the joint angles of a humanoid robot to generate the humanoid pose similar to the human pose. This interface is implemented as the plugin of Choreonoid.

Mitek Anchor–Augmented Static Facial Suspension

To evaluate the efficacy of the Mitek minianchor suture system as an adjunctive procedure for static facial suspension and assess its ability to improve functional and aesthetic outcomes in patients with complete facial paralyses. A retrospective review of 5 patients who underwent adjunctive static facial suspension with the Mitek GII minianchor suture system at a tertiary care academic center. A bone anchor in the maxilla connected to 2 suture arms was used. One suture arm elevated the oral commissure on the paralyzed side. The second suture arm adjusted for upper lip segment asymmetry and Cupid's bow deviation, or lateralized the external nasal valve. Outcome was evaluated by photodocumentation and clinical examination. There was an average of 76.9% improvement in vertical symmetry of the oral commissure (range, 43%-100%), an average of 65.0% improvement in horizontal deviation of the Cupid's bow (range, 50%-100%), and an average of 85.3% improvement in symmetry of upper lip segments (range, 67%-100%). All patients experienced improvement in oral-nasal competency. The Mitek minianchor suture system is a safe and effective adjunctive method to improve facial symmetry and oral-nasal competency in patients with facial paralyses. Its multivector design approximates the ideal positions of key anatomical points of the midface during static facial suspension surgery.

Mitek Anchor–Augmented Static Facial Suspension

To evaluate the efficacy of the Mitek minianchor suture system as an adjunctive procedure for static facial suspension and assess its ability to improve functional and aesthetic outcomes in patients with complete facial paralyses. A retrospective review of 5 patients who underwent adjunctive static facial suspension with the Mitek GII minianchor suture system at a tertiary care academic center. A bone anchor in the maxilla connected to 2 suture arms was used. One suture arm elevated the oral commissure on the paralyzed side. The second suture arm adjusted for upper lip segment asymmetry and Cupid's bow deviation, or lateralized the external nasal valve. Outcome was evaluated by photodocumentation and clinical examination. There was an average of 76.9% improvement in vertical symmetry of the oral commissure (range, 43%-100%), an average of 65.0% improvement in horizontal deviation of the Cupid's bow (range, 50%-100%), and an average of 85.3% improvement in symmetry of upper lip segments (range, 67%-100%). All patients experienced improvement in oral-nasal competency. The Mitek minianchor suture system is a safe and effective adjunctive method to improve facial symmetry and oral-nasal competency in patients with facial paralyses. Its multivector design approximates the ideal positions of key anatomical points of the midface during static facial suspension surgery.

Crawler Cranes Modeling Method Based on OGRE

According to the crawler crane’s structure and the scene management principle of OGRE (Object-Oriented Graphics Rendering Engine), this paper presents a new 3D parameterized modeling method of crawler cranes. By integrating the interactive modeling using 3DSMAX and the geometrical graphics modeling, the new method can solve the problem about rendering speed, vivid and parameterized models and how to obtain the position of key points. A case is described in order to demonstrate the use of the developed method and to illustrate its reliability.

MARKERLESS HUMAN MOTION TRACKING FROM A SINGLE CAMERA USING INTERVAL PARTICLE FILTERING

In this paper we present a new approach for marker less human motion capture from conventional camera feeds. The aim of our study is to recover 3D positions of key points of the body that can serve for gait analysis. Our approach is based on foreground extraction, an articulated body model and particle filters. In order to be generic and simple, no restrictive dynamic modeling was used. A new modified particle-filtering algorithm was introduced. It is used efficiently to search the model configurations space. This new algorithm, which we call Interval Particle Filtering, reorganizes the configurations search space in an optimal deterministic way and proved to be efficient in tracking natural human movement. Results for human motion capture from a single camera are presented and compared to results obtained from a marker based system. The system proved to be able to track motion successfully even in partial occlusions and even outdoors.

Dynamic cell structures for the evaluation of keypoints in facial images.

In this contribution Dynamic Cell Structures (DCS network) are applied to classify local image structures at particular facial landmarks. The facial landmarks such as the corners of the eyes or intersections of the iris with the eyelid are computed in advance by a combined model and data driven sequential search strategy. To reduce the detection error after the processing of the sequential search strategy, the computed image positions are verified applying a DCS network. The DCS network is trained by supervised learning with feature vectors which encode spatially arranged edge and structural information at the keypoint position considered. The model driven localization as well as the data driven verification are based on steerable filters, which build a representation comparable with one provided by a receptive field in the human visual system. We apply a DCS based classifier because of its ability to grasp the topological structure of complex input spaces and because it has proved successful in a number of other classification tasks. In our experiments the average error resulting from false positive classifications is less than 1%.

Synthesising continuous-tone caricatures

A series of tools has been developed to synthesize full colour photographic quality facial caricatures. Real faces, photographs or single-frame video films are frame-grabbed and the position of (186) key points around external and internal features are interactively defined. Following procedures of Brennan 7,8 the 2D Cartesian coordinates of each feature point are compared with those of an average for many faces; differences in positions are exaggerated (to produce caricature data sets) or reduced (giving ‘anticaricature’ data sets). Line-drawing representations of faces and their caricatures are formed by linking appropriate feature points. To produce continuous-tone caricatures, the original and caricature images are resolved into (340) adjacent triangular tessellations. Rendering of the final image is accomplished by mapping the pixel values (grey-scale or colour) from each source triangle in the original image to the corresponding triangle in the caricature image. The caricatures not only provide amusement, but can be used as stimuli in psychological investigations into the representation of faces in human memory.