Flavio Prieto

Automatic Lip-Contour Extraction and Mouth-Structure Segmentation in Images

Lip-contour extraction has great potential for human-machine interface and communication systems, but most existing techniques are inappropriate for changing poses, malformations, or whole-mouth descriptions. A new mouth-structure segmentation methodology uses pixel color classification, segmentation refinement, and fitted region-of-interest clipping to improve the speed and accuracy of mouth-structure segmentation using standard database images.

Automatic Hole-Filling of Triangular Meshes Using Local Radial Basis Function

Creating models of real objects is a complex task for which the use of traditional modeling techniques has proven to be difficult. To solve some of these problems, laser rangefinders are frequently used to sample an objects surface from several viewpoints resulting in a set of range images that are registered and integrated into a final triangulated model. In practice, due to surface reflectance properties, occlusions and accessibility limitations, certain areas of the objects surface are usually not sampled, leaving holes which create undesirable artifacts in the integrated model. In this paper, we present a novel algorithm for the automatic hole-filling of triangulated models. The algorithm starts by locating hole boundary regions. A hole consists of a closed path of edges of boundary triangles that have at least an edge, which is not shared with any other triangle. The edge of the hole is then fitted with a b-spline where the average variation of the torsion of the b-spline approximation is calculated. Using a simple threshold of the average variation of the torsion along the edge, one can automatically classify real holes from man-made holes. Following this classification process, we then use an automated version of a radial basis function interpolator to fill the inside of the hole using neighboring edges. Excellent experimental results are presented.

CAD-based range sensor placement for optimum 3D data acquisition

The use of laser range sensor allows a very significant improvement in acquisition speed but does not equal the accuracy obtained with a coordinate measuring machine. In order to obtain a quality control close to that obtained in metrology we suggest improving the accuracy of the depth measurements by positioning the sensors head according to a strategy for optimum 3D data acquisition. We propose such a strategy to automatically produce a sensing plan for completely and accurately acquiring the geometry of a surface or of the whole piece whenever possible. The system requires the exact position and orientation of the part and its CAD model in IGES format. There is no limitation regarding the shape of the part to be digitized. An auto-synchronized range sensor developed at the NRCC was used, and for this sensor, the accuracy of the 3D measured points is a function of the distance and of the incident angle relative to the surface. Our strategy guarantees that the viewpoints found meet the best accuracy conditions in the scanning process.

Fully automatic expression-invariant face correspondence

We consider the problem of computing accurate point-to-point correspondences among a set of human face scans with varying expressions. Our fully automatic approach does not require any manually placed markers on the scan. Instead, the approach learns the locations of a set of landmarks present in a database and uses this knowledge to automatically predict the locations of these landmarks on a newly available scan. The predicted landmarks are then used to compute point-to-point correspondences between a template model and the newly available scan. To accurately fit the expression of the template to the expression of the scan, we use as template a blendshape model. Our algorithm was tested on a database of human faces of different ethnic groups with strongly varying expressions. Experimental results show that the obtained point-to-point correspondence is both highly accurate and consistent for most of the tested 3D face models.

A multi-user remote academic laboratory system

This article describes the development, implementation and preliminary operation assessment of Multiuser Network Architecture to integrate a number of Remote Academic Laboratories for educational purposes on automatic control. Through the Internet, real processes or physical experiments conducted at the control engineering laboratories of four universities are remotely operated. Through an internet connection to the Manager Administration Server, a remote experiment to design and test a modeling and control algorithm can be performed. The suggested Network Architecture is based on the Singlet-Server model and uses a database server to record important information that helps create a new remote experiment, including a Graphical User Interface (Applet) developed with Easy Java Simulation, which allows the simple integration of new processes to the Manager Administrator. Results of a real-physical-process remote manipulation through the proposed network architecture are presented as well as results of an academic pilot test conducted to measure functional aspects related to the operation of the remote system when carrying out remote-laboratory work.

A CAD-based 3D data acquisition strategy for inspection

Abstract.The use of a laser range sensor in the 3D digitalization process allows significant improvement in acquisition speed and in 3D measurement point density. However, if we want to use these 3D data in applications that require data with a high degree of accuracy like inspection tasks, it is mandatory that the 3D points be acquired under the best conditions of accuracy. During 3D capture of a part, several sources of error can alter the measured values. Thus we must find and model the most important parameters affecting the accuracy of the range sensor. This error model, along with the CAD model of the part, is used to produce a sensing plan to completely and accurately acquire the geometry of the part. The sensing plan is comprised of the set of viewpoints that defines the exact position and orientation of the camera relative to the part. There is no limitation with regard to the shape of the part to be digitalized. An autosynchronized range sensor fixed on a coordinate measuring machine was used. For this sensor, the accuracy of the 3D measured points is a function of the distance and of the angle of incidence relative to the surface. The strategy proposed to find the acquisition plan guarantees that the viewpoints meet the best accuracy conditions in the scanning process, solving the occlusion problems. It was found that the 3D data acquired by using the proposed strategy are around 30% more accurate than the 3D data obtained in a standard acquisition.

Mouth gesture and voice command based robot command interface

In this paper we present a voice command and mouth gesture based robot command interface which is capable of controlling three degrees of freedom. The gesture set was designed in order to avoid head rotation and translation, and thus relying solely in mouth movements. Mouth segmentation is performed by using the normalized a* component, as in [1]. The gesture detection process is carried out by a Gaussian Mixture Model (GMM) based classifier. After that, a state machine stabilizes the system response by restricting the number of possible movements depending on the initial state. Voice commands are modeled using a Hidden Markov Model (HMM) isolated word recognition scheme. The interface was designed taking into account the specific pose restrictions found in the DaVinci Assisted Surgery command console.

A Hole-Filling Algorithm for Triangular Meshes Using Local Radial Basis Function

Creating models of real objects is a complex task for which the use of traditional modeling techniques has proven to be difficult. To solve some of the problem encountered, laser rangefinders are frequently used to sample an object ́s surface from several viewpoints resulting in a set of range images that are registered and integrated into a final triangulated model. In practice, due to surface reflectance properties, occlusions and accessibility limitations, certain areas of the object ́s surface are not sampled leaving holes which create undesirable artifacts in the integrated model. In this paper, we present a novel algorithm for the automatic hole–filling of triangulated models. The algorithm starts by locating hole boundary regions. A hole consists of a closed path of edges of boundary triangles that have at least an edge, which is not shared with any other triangle. The edge of the hole is then fitted with a b-spline where the average variation of the torsion of the b-spline approximation is calculated. Using a simple threshold of the average variation of the torsion along the edge, one can automatically classify real holes from man-made holes. Following this classification process, we then use an automated version of a radial basis function interpolator to fill the inside of the hole using neighboring edges. Excellent experimental results are presented. 412 J. Branch et al.

Literature review of acoustic and ultrasonic tomography in standing trees

Key messageHigh-resolution imaging is possible if high-frequency sensors are used together with a signal-processing and inversion algorithm that is well suited to a low signal-to-noise ratio and the effect of wood anisotropy.AbstractWood is a biological growth medium, and given that standing trees adapt themselves in their growth to environmental conditions, their material properties vary with age. These changes result in variations that are far more complex than anisotropy. Wood quality and intraspecific variability can thus be studied to gain an understanding of the development mechanisms of trees, and this can be useful for clonal selection and the management of tree communities. A number of techniques are available to determine wood properties in standing trees, but the signal-processing approaches currently used are not always robust and do not always provide the image resolution needed in the particular cases of acoustic or ultrasonic tomography. This review paper thus aims to present important aspects that should be taken into account when using tomography techniques and addresses a number of open problems. A brief review of current non-destructive wood imaging techniques is initially presented followed by a comparison of the protocols, methods and models used in acoustic and ultrasonic tomography. The devices cited were studied in terms of measurement systems and signal processing. The analysis aimed to highlight and analyze the advantages and disadvantages of each device and describe challenges and trends. The effect of various parameters is discussed: frequency, signal-to-noise ratio, number of sensors and inversion algorithm. General conclusions are then drawn in relation to future signal-processing work in the acoustic and ultrasonic tomography of standing trees.

On-line 3-D inspection of deformable parts using FEM trained radial basis functions

Usually, the inspection of manufactured parts in industry is done by performing a comparison between a reference model, such as a CAD model, and 3-D measurements taken on the surface of the part using systems like coordinate measuring machines, photogrammetry, or laser scanners. This process involves first the alignment of both reference models and measured dataset to determine if the part fulfill the required tolerances specified by the designer. However, due to new flexible materials as well as structural shape many parts are remarkably flexible. In those cases, it is necessary to carry out a fixturing process in order to constrain the part to its nominal shape. This fixturing process is time consuming and hard to automate. This paper presents a new method for the alignment of deformable parts that do not require any fixturing step. The proposed approach uses a finite element method to obtain a physical deformation of the original CAD model, and radial basis functions to approximate this deformation faster and in real-time, opening the door to on-line inspection of deformable parts. Experimental data are presented.