Gianluca Percoco

Assembly and Disassembly Planning by using Fuzzy Logic & Genetic Algorithms:

The authors propose the implementation of hybrid Fuzzy Logic-Genetic Algorithm (FL-GA) methodology to plan the automatic assembly and disassembly sequence of products. The GA-Fuzzy Logic approach is implemented onto two levels. The first level of hybridization consists of the development of a Fuzzy controller for the parameters of an assembly or disassembly planner based on GAs. This controller acts on mutation probability and crossover rate in order to adapt their values dynamically while the algorithm runs. The second level consists of the identification of the optimal assembly or disassembly sequence by a Fuzzy function, in order to obtain a closer control of the technological knowledge of the assembly/disassembly process. Two case studies were analyzed in order to test the efficiency of the Fuzzy-GA methodologies.

Reverse engineering techniques applied to a human skull, for CAD 3D reconstruction and physical replication by rapid prototyping

The production of a copy of an existing object of complex shape is one of the typical applications of the integration between two modern computer-based technologies, reverse engineering (RE) and rapid prototyping (RP). The method is extremely versatile and can be used in various applicative domains (e.g. replacement of anatomical parts with artificial prostheses, replication of skeletal remains). Two different acquisition techniques of images of a skull, by laser and by CT scan, were compared to ascertain which enabled more accurate reproduction of the original specimen. The skull was chosen due to it being the body part most often used in medico-legal investigations (for personal identification, skull-photo superimposition techniques, forensic art, etc). Comparison between the copy and the original yielded satisfactory results for both techniques. However, CT scanning demonstrated some advantages over the laser technique, as it provided a cleaner point cloud, enabling shorter pre-reproduction processing times, as well as data on the internal parts, which resulted in the reproduction of a more faithful copy.

Automated landmark extraction for orthodontic measurement of faces using the 3-camera photogrammetry methodology.

Objectives: To set up a three-dimensional photogrammetric scanning system for precise landmark measurements, without any physical contact, using a low-cost and noninvasive digital photogrammetric solution, for supporting several necessity in clinical orthodontics and/or surgery diagnosis. Materials and Methods: Thirty coded targets were directly applied onto the subjects face on the soft tissue landmarks, and then, 3 simultaneous photos were acquired using photogrammetry, at room light conditions. For comparison, a dummy head was digitized both with a photogrammetric technique and with the laser scanner Minolta Vivid 910i (Konica Minolta, Tokyo, Japan). Results: The precise measurement of the landmarks is ranged between 0.017 and 0.029 mm. The system automatically measures spatial position of face landmarks, from which distances and angles can be obtained. The facial measurements were compared with those done using laser scanning and manual caliper. The adopted method gives higher precision than the others (0.022-mm mean value on points and 0.038-mm mean value on linear distances on a dummy head), is simple, and can be used easily as a standard routine. Conclusions: The study demonstrated the validity of photogrammetry for accurate digitization of human face landmarks. This research points out the potential of this low-cost photogrammetry approach for medical digitization.

Coded targets and hybrid grids for photogrammetric 3D digitisation of human faces

The three-dimensional (3D) measure of the human body is currently performed using mostly optical technologies. One of the most cost effective non-contact techniques is photogrammetry; its main disadvantage is the lack of automation because the correspondences between the same points in different images must be taken manually. In this paper the authors present a properly designed low-cost photogrammetric system for 3D scanning of human faces. Results are compared projecting onto the faces patterns composed by coded targets and mixed coded-uncoded targets.

Accurate facial morphologic measurements using a 3-camera photogrammetric method.

Objectives:A new, low-cost photogrammetric method has been developed for facial morphometry applications. To evaluate the system, tests for the measurement and comparison of three-dimensional virtual faces were carried out in different subjects. Materials and Methods:Twenty adult white Italian subjects, 10 men and 10 women, of ages ranging from 23 to 37 years, were included in this study. Three cameras were finely calibrated, and the point precision vector length was calculated, together with the quality parameters. For each subject, 3 different acquisitions were performed. A tessellated surface was obtained from each point cloud. The comparison was made by aligning three-dimensional information from different models. Differences between 2 different models were estimated by analysis of the distances. Results:For the cases analyzed, the mean point precision overall root-mean-square vector length was 0.07 mm, with a SD of 0.027 mm. The results are reported for the systems capability of discriminating between the faces of different people. Results of comparisons between facial models of a single person were compared with those of comparisons between different subjects. Students t-test revealed that the system was able to discriminate among different people, with a P > 95%. Two sex subgroups were formed: the mean error between subgroups ranged from 1.65 to 3.43 mm, and the mean ranged from 1.76 to 2.72 mm. Conclusions:The experiments confirmed the capabilities and the accuracy of the proposed photogrammetric system. Facial comparison was performed by analysis of distances on three-dimensional virtual models.

Digital Photogrammetry for Facial Recognition

In this paper, the authors present a biometric low-cost 3D acquisition system, based on a digital photogrammetry technique. The aim of the work is to analyze the suitability of this system for facial recognition purposes. The facial data of a set of 20 people were acquired with the photogrammetric system developed by the authors, and different CAD 3D models were reconstructed for each person. The results are quantified by aligning the models and calculating mean distances and standard deviations between them using two different methods. The former considers the entire face; the latter is based on a few fiducial points of the face.

An artificial intelligence approach to registration of free-form shapes

Abstract Registration, defined as the process of matching geometric entities, is performed when multiple scanned data sets must be aligned or when an existing model must match digitized point clouds. This process is crucial in several applications such as Reverse Engineering, CAD-based inspection and computer vision. The goal of this process is the computation of the optimal rigid transformation for the alignment of several sets of geometric entities (points and/or surfaces). Registration is generally performed by using a two-step procedure necessary to realize coarse and fine alignments. Human intervention is normally required for coarse registration while fine registration is usually a semi-automatic procedure. Consequently alignment is not usually a single step automatic operation and is also affect by errors. In this paper the authors propose a hybrid approach for automatic registration applied to free-form shapes. This hybrid approach employs a asynchronous data communication between an Artificial Neural Network and Genetic Algorithms. The Neural Network performs the coarse alignment giving an initial solution for the registration operation which is then performed by Genetic Algorithms to minimize error deviations between geometrical entities. Several case studies have been investigated in order to validate the proposed approach.

Is principal component analysis an effective tool to predict face attractiveness? A contribution based on real 3D faces of highly selected attractive women, scanned with stereophotogrammetry.

Abstract In the literature, several papers report studies on mathematical models used to describe facial features and to predict female facial beauty based on 3D human face data. Many authors have proposed the principal component analysis (PCA) method that permits modeling of the entire human face using a limited number of parameters. In some cases, these models have been correlated with beauty classifications, obtaining good attractiveness predictability using wrapped 2D or 3D models. To verify these results, in this paper, the authors conducted a three-dimensional digitization study of 66 very attractive female subjects using a computerized noninvasive tool known as 3D digital photogrammetry. The sample consisted of the 64 contestants of the final phase of the Miss Italy 2010 beauty contest, plus the two highest ranked contestants in the 2009 competition. PCA was conducted on this real faces sample to verify if there is a correlation between ranking and the principal components of the face models. There was no correlation and therefore, this hypothesis is not confirmed for our sample. Considering that the results of the contest are not only solely a function of facial attractiveness, but undoubtedly are significantly impacted by it, the authors based on their experience and real faces conclude that PCA analysis is not a valid prediction tool for attractiveness. The database of the features belonging to the sample analyzed are downloadable online and further contributions are welcome.

Three-dimensional methodology for photogrammetric acquisition of the soft tissues of the face: a new clinical-instrumental protocol

BackgroundThe objective of this study is to define an acquisition protocol that is clear, precise, repeatable, simple, fast and that is useful for analysis of the anthropometric characteristics of the soft tissue of the face.MethodsThe analysis was carried out according to a new clinical-instrumental protocol that comprises four distinct phases: (1) setup of portable equipment in the space in which field analysis will be performed, (2) preparation of the subject and spatial positioning, (3) scanning of the subject with different facial expressions, and (4) treatment and processing of data. The protocol was tested on a sample comprising 66 female subjects (64 Caucasian, 1 Ethiopian, and 1 Brazilian) who were the finalists of an Italian national beauty contest in 2010. To illustrate the potential of the method, we report here the measurements and full analysis that were carried out on the facial model of one of the subjects who was scanned.ResultsThis new protocol for the acquisition of faces is shown to be fast (phase 1, about 1 h; phase 2, about 1.5 min; phase 3, about 1.5 min; phase 4, about 15 min), simple (phases 1 to 3 requiring a short operator training period; only phase 4 requires expert operators), repeatable (with direct palpation of anatomical landmarks and marking of their positions on the face, the problem of identification of these same landmarks on the digital model is solved), reliable and precise (average precision of measurements, 0.5 to 0.6 mm over the entire surface of the face).ConclusionsThis standardization allows the mapping of the subjects to be carried out following the same conditions in a reliable and fast process for all of the subjects scanned.

Noninvasive computerized scanning method for the correlation between the facial soft and hard tissues for an integrated three-dimensional anthropometry and cephalometry.

ObjectivesThe article describes a new methodology to scan and integrate facial soft tissue surface with dental hard tissue models in a three-dimensional (3D) virtual environment, for a novel diagnostic approach.The facial and the dental scans can be acquired using any optical scanning systems: the models are then aligned and integrated to obtain a full virtual navigable representation of the head of the patient. MethodsIn this article, we report in detail and further implemented a method for integrating 3D digital cast models into a 3D facial image, to visualize the anatomic position of the dentition. This system uses several 3D technologies to scan and digitize, integrating them with traditional dentistry records. The acquisitions were mainly performed using photogrammetric scanners, suitable for clinics or hospitals, able to obtain high mesh resolution and optimal surface texture for the photorealistic rendering of the face. To increase the quality and the resolution of the photogrammetric scanning of the dental elements, the authors propose a new technique to enhance the texture of the dental surface. ResultsThree examples of the application of the proposed procedure are reported in this article, using first laser scanning and photogrammetry and then only photogrammetry. Using cheek retractors, it is possible to scan directly a great number of dental elements. The final results are good navigable 3D models that integrate facial soft tissue and dental hard tissues. The method is characterized by the complete absence of ionizing radiation, portability and simplicity, fast acquisition, easy alignment of the 3D models, and wide angle of view of the scanner. ConclusionsThis method is completely noninvasive and can be repeated any time the physician needs new clinical records. The 3D virtual model is a precise representation both of the soft and the hard tissue scanned, and it is possible to make any dimensional measure directly in the virtual space, for a full integrated 3D anthropometry and cephalometry. Moreover, the authors propose a method completely based on close-range photogrammetric scanning, able to detect facial and dental surfaces, and reducing the time, the complexity, and the cost of the scanning operations and the numerical elaboration.