Correspondences between 3D keypoints generated by matching local descriptors are a key step in 3D computer vision and graphic applications. Learned descriptors are rapidly evolving and outperforming the classical handcrafted approaches in the field. Yet, to learn effective representations they require supervision through labeled data, which are cumbersome and time-consuming to obtain. Unsupervised alternatives exist, but they lag in performance. Moreover, invariance to viewpoint changes is attained either by relying on data augmentation, which is prone to degrading upon generalization on unseen datasets, or by learning from handcrafted representations of the input which are already rotation invariant but whose effectiveness at training time may significantly affect the learned descriptor. We show how learning an equivariant 3D local descriptor instead of an invariant one can overcome both issues. LEAD (Local EquivAriant Descriptor) combines Spherical CNNs to learn an equivariant representation together with plane-folding decoders to learn without supervision. Through extensive experiments on standard surface registration datasets, we show how our proposal outperforms existing unsupervised methods by a large margin and achieves competitive results against the supervised approaches, especially in the practically very relevant scenario of transfer learning.

Several student drivers have a high level of stress and may need assistance of a specialist, such as a psychologist, in order to enhance their driving skills or even be able to drive on their own. In this context, it is advantageous that such specialists possess resources to deeply analyze and understand the student reactions when submitted to practical driving activities. The literature already include works focusing on automatic detection of driver stress and methods for assisting motorists in real time. However, there is an open research gap regarding the production of reports and summaries about driving activities that can be useful for students behavior analysis and treatment. To this end, we propose an approach for analyzing and summarizing information about driver stress based on their behavior. The approach is supported by a computational tool that allows to view different types of information about the driver under three perspectives: i) videos of the driving activities; ii) reports of behavior analysis; and iii) summaries of relevant actions. A dataset with videos, heart rate and geographic location of driving activities developed by student drivers was produced. The dataset is initially labeled, then the discrete label values are transformed to continuous values to improve visualization and summarization. The approach was evaluated qualitatively by a psychologist and driving instructors. The proposed approach helps a professional to quickly understand the drivers' profile, interpreting the causes of the drivers' reactions, thus providing more accurate assistance.

Technology can support people with disabilities to participate in social and economic life. Using relevant Human-Computer Interaction, as obtained through Intelligent User Interfaces, people with motor and speech impairments may be able to communicate in different ways. Augmentative and Alternative Communication supported by Computer Vision systems can benefit from the recognition of users' remaining functional motions as an alternative interaction design approach. Based on a methodology in which gestures and their meanings are created and configured by users and their caregivers, we developed a Computer Vision system, named PGCA, that employs machine learning techniques to create personalized gestural interaction as an Assistive Technology resource for communication purposes. Using a low-cost approach, PGCA has been experienced with students with motor and speech impairments to create personalized gesture datasets and to identify improvements for the system. This paper presents an experiment carried with the target audience using a game-based approach where three students used PGCA to interact with communication boards and to play a game. The system was evaluated by special education professionals using the System Usability Scale and was considered suitable for its purpose. Results from the experiment suggest the technical feasibility for the methodology and for the system, also adding knowledge about the interaction process of disabled people with a game.

Defining and reliably finding a canonical orientation for 3D surfaces is key to many Computer Vision and Robotics applications. This task is commonly addressed by handcrafted algorithms exploiting geometric cues deemed as distinctive and robust by the designer. Yet, one might conjecture that humans learn the notion of the inherent orientation of 3D objects from experience and that machines may do so alike. In this work, we show the feasibility of learning a robust canonical orientation for surfaces represented as point clouds. Based on the observation that the quintessential property of a canonical orientation is equivariance to 3D rotations, we propose to employ Spherical CNNs, a recently introduced machinery that can learn equivariant representations defined on the Special Ortoghonal group SO(3). Specifically, spherical correlations compute feature maps whose elements define 3D rotations. Our method learns such feature maps from raw data by a self-supervised training procedure and robustly selects a rotation to transform the input point cloud into a learned canonical orientation. Thereby, we realize the first end-to-end learning approach to define and extract the canonical orientation of 3D shapes, which we aptly dub Compass. Experiments on several public datasets prove its effectiveness at orienting local surface patches as well as whole objects.

This paper presents a multi-label Action Unit (AU) detection method applied on multi-pose facial images. Action Unit detection on multiple head poses is an issue that robust AU detectors must deal with, as it is uncommon for a person to maintain always the same pose when displaying facial expressions. To this end, this work proposes a region learning approach, that dynamically creates regions of interest inside a convolutional neural network (CNN) using facial landmark points. The dynamic region learning (DRL) ensures that each AU is in the center of the region, and also follows the head pose movement. The DRL is built on top of the VGG-Face network, and transfer-learning is used to start the training. The experiments were conducted on the Facial Expression Recognition and Analysis Challenge (FERA 2017) database, which contains nine different head poses. The results show that the dynamic region learning is able to adapt to the nine poses in the database, improving the state-of-the-art with an an average F1-score of 0.582.

This paper proposes a novel methodology for biometric identification of individuals using level-3 features (pores), extracted from 3D fingerprint images obtained through Optical Coherence Tomography (OCT). OCT fingerprint images contain detailed 3D information from both the dermis and the epidermis skin layers of fingertips. Our approach first fetches and extracts pores around minutiae from the 3D fingerprint data, creating small structures called pore clouds. The correspondence of existent pore clouds are then verified for all the three possible fingerprint matching: dermis-dermis, epidermis-epidermis, and dermis-epidermis. To this end, three different measures are extracted and compared: the Hausdorff distance, the Surface Interpenetration Measure and the Root Mean Square Error. Experiments using 518 pore clouds achieved recognition rates of 99.19% for Rank-1 with EER (Equal Error Rate) of 0.72%. From our best knowledge, this is the first time the identification of individuals using only 3D information from pores is explored.

This paper presents an unified convolutional neural network (CNN), named AUMPNet, to perform both Action Units (AUs) detection and intensity estimation on facial images with multiple poses. Although there are a variety of methods in the literature designed for facial expression analysis, only few of them can handle head pose variations. Therefore, it is essential to develop new models to work on non-frontal face images, for instance, those obtained from unconstrained environments. In order to cope with problems raised by pose variations, an unique CNN, based on region and multitask learning, is proposed for both AU detection and intensity estimation tasks. Also, the available head pose information was added to the multitask loss as a constraint to the network optimization, pushing the network towards learning better representations. As opposed to current approaches that require ad hoc models for every single AU in each task, the proposed network simultaneously learns AU occurrence and intensity levels for all AUs. The AUMPNet was evaluated on an extended version of the BP4D-Spontaneous database, which was synthesized into nine different head poses and made available to FG 2017 Facial Expression Recognition and Analysis Challenge (FERA 2017) participants. The achieved results surpass the FERA 2017 baseline, using the challenge metrics, for AU detection by 0.054 in F1-score and 0.182 in ICC(3, 1) for intensity estimation.

We propose a novel, non-invasive method for identification of human fingerprints obtained from 3D images of the dermis and epidermis. Using images obtained with optical coherence tomography, we compute the Gaussian (K) and mean (H) curvature values of the dermatoglyphics of dermal and epidermal images from volunteers' fingers, which are then converted into curvature-type and KH maps. Next, the regions of maps located around the minutiae are matched based on their normalized cross correlation. To the best of our knowledge, this is the first work to explore the use of KH maps of the finger dermatoglyphics and 3D images of the dermis to support fingerprint identification. The reliability of 3D dermal fingerprints is further explored by comparing unrolled 2D fingerprints extracted from 3D dermal point clouds against a 2D fingerprint test database obtained using a 2D commercial scanner and software. Finally, an experiment is performed to illustrate the robustness of the dermal fingerprint to mild alterations of the epidermis.

Using 3D printers for manufacturing objects has become an easy, low cost process in a variety of emerging applications, such as assistive technologies. To this end, it may be necessary to split the object in smaller parts, providing that they precisely connect. This work presents a framework to build objects with connecting parts, which is divided in main three steps: 3D reconstruction; mesh segmentation; and 3D printing. The framework includes a new method for 3D model segmentation, which allows the generation of connecting parts from cutting planes. The segmentation process uses a binary space partitioning tree to represent cut regions of the mesh obtained from the scanned object. To close these cut regions and, at the same time, to consider constraints defined by the added connectors, it is used a Constrained Delaunay Triangulation. Thus, it is guaranteed a set of closed meshes for 3D printing. Experimental results show the application of this framework in two different areas: assistive technologies, by creating anatomically adjustable orthoses; and cultural heritage, where parts of scanned sculptures can be printed for preservation and restoration activities.

This work presents a new data-driven progressive compression method of colored 3D meshes. The proposed method improves the visual quality of the progressive compression by using texture mapping. Additionally, a cubic cell complex is computed to drive the compression process in order to allow the intermediate meshes to share a common colored texture. The experimental results show that our approach based on color textures improves the visual quality of low resolution intermediate meshes as compared to traditional methods based on color per vertex. The method is well suited for remote visualization scenarios where the visual appearance is as important as the compact representation of 3D meshes.

We present a novel method for newborn authentication that matches key points in different interdigital regions from palm prints or footprints. Then, the method hierarchically combines the scores for authentication. We also present a novel pore detector for key point extraction, named Dynamic Pore Filtering (DPF), that does not rely on expensive processing techniques and adapts itself to different sizes and shapes of pores. We evaluated our pore detector using four different datasets. The obtained results of the DPF when using newborn dermatoglyphic patterns (2400ppi) are comparable to the state-of-the-art results for adult fingerprint images with 1200ppi. For authentication, we used four datasets acquired by two different sensors, achieving true acceptance rates of 91.53% and 93.72% for palm prints and footprints, respectively, with a false acceptance rate of 0%. We also compared our results to our previous approach on newborn identification, and we considerably outperformed its results, increasing the true acceptance rate from 71% to 98%.

Abstract This work presents a new representation used as a rendering primitive of surfaces. Our representation is defined by an arbitrary cubic cell complex: a projection‐based parameterization domain for surfaces where geometry and appearance information are stored as tile textures. This representation is used by our ray casting rendering algorithm called projection mapping, which can be used for rendering geometry and appearance details of surfaces from arbitrary viewpoints. The projection mapping algorithm uses a fragment shader based on linear and binary searches of the relief mapping algorithm. Instead of traditionally rendering the surface, only front faces of our rendering primitive (our arbitrary cubic cell complex) are drawn, and geometry and appearance details of the surface are rendered back by using projection mapping. Alternatively, another method is proposed for mapping appearance information on complex surfaces using our arbitrary cubic cell complexes. In this case, instead of reconstructing the geometry as in projection mapping, the original mesh of a surface is directly passed to the rendering algorithm. This algorithm is applied in the texture mapping of cultural heritage sculptures.

This work proposes to improve a traditional 3D reconstruction pipeline by combining it with two techniques: A realtime 3D modelling system to give a visual feedback in acquisition stage; and a super-resolution spatio-temporal filter to improve the depth data. We use noisy RGBD images from an inaccurate real-time depth sensor device in the whole process and later replace the color with data from a digital camera to generate realistic texture. Recently, several reconstruction approaches were presented acknowledging the potential of such real-time devices. However, those systems alone fail to retrieve small geometric characteristics from the object. The result of our experiments show a model with a considerable level of details, unexpected from low-quality RGBD images. We aim to use our pipeline to reconstruct objects with cultural value and add them to the virtual museum's database for visualization purpose.

In this work, we automate a 3D reconstruction pipeline using two SURF-based approaches. The first approach uses SURF correspondences to pre-align multiple 3D scans from a same object without requiring manual labor. The second approach uses SURF correspondences to calibrate high resolution color images to 3D scans in order to improve the texture quality of the final 3D model. Both approaches succeeded in more than 95% of the test cases and were able to automatically identify incorrect results. Our pipeline has been widely used in several projects of cultural heritage and the proposed improvements are very important to allow its use in large collections. The proposed approaches were favorably compared to other methods and have been successfully applied in cultural heritage preservation of sculptures located in a UNESCO World Heritage site.

In this work we evaluate the influence of pose variation on 3D face reconstruction from a single image. To this end, we present a 3D reconstruction method that combines a fitting technique and a sparse 3D deformable model to estimate the 3D information of 2D images with large pose variations. For our experiments, we synthetically created 2D images by rendering 3D models from the BU-3DFE database in different points of view. Thus, we have a precise ground truth that allows performing a quantitative analysis of the reconstruction accuracy. Our experimental results show that the reconstruction achieves the highest accuracy when using half-frontal face images, and is also more robust to noise and incorrect facial landmarks positioning.

We present some results on newborn identification through high-resolution images of palmar surfaces. To our knowledge, there is no biometric system currently available that can be effectively used for newborn identification. The manual procedure of capturing inked footprints in practice for this purpose is limited for use inside hospitals and is not an effective solution for identification purposes. The use of friction ridge patterns on the hands of newborns is challenging due to both the small size of newborn's papillary ridges, which are, on average, 2.5 to 3 times smaller than the ridges in adult fingerprints, and their fragility, making them amenable to deformation. The proposed palmprint based automatic system for newborn identification is relatively easy to use and shows the feasibility of this approach. Experiments were performed on images collected from 250 newborns at the University Hospital (Universidade Federal do Parana). An image acquisition protocol was developed in order to collect suitable images. When considering the good quality palmar images, the results show that the pro- posed approach is promising.

This work presents a new component of a complete pipeline for 3D digital preservation of cultural heritage. We propose a method for parameterizing and preserving the appearance of simplified models. The method automatically preserves the appearance, passing surface properties such as normals or colors from a high-resolution model to its simplified version. Surface properties of the original model are projected onto a quadrilateral domain defined by a collection of cubic cells, and then these properties are stored into a 2D texture. Later in the rendering process, high-resolution surface properties are mapped from the texture onto the simplified model. The main contributions are: a parameterization for arbitrary surfaces based on cubic cells, an automatic generation of texture coordinates using only the surface of a model, and a method for preserving the appearance of simplified models.

We present a scale-invariant face detection approach based on boosted cascade classifiers using range images as input. The detector was developed to be employed as a preliminary stage for any real-time 3D face recognition system. The required computation time for this task was considerably reduced by eliminating the need for scanning an input image in multiple scales. Our experiments were performed using two well-known databases, and the proposed approach was favorably compared against a state-of-the-art face detection approach. We achieved a detection rate of 99.9% with only 0.2% of the images presenting false detections. We also evaluated the detector performance in face images presenting large pose variations and obtained detection rates as high as when using frontal face images.

We present a 3D face reconstruction system that takes as input either one single view or several different views. Given a facial image, we first classify the facial pose into one of five predefined poses, then detect two anchor points that are then used to detect a set of predefined facial landmarks. Based on these initial steps, for a single view we apply a warping process using a generic 3D face model to build a 3D face. For multiple views, we apply sparse bundle adjustment to reconstruct 3D landmarks which are used to deform the generic 3D face model. Experimental results on the Color FERET and CMU multi-PIE databases confirm our framework is effective in creating realistic 3D face models that can be used in many computer vision applications, such as 3D face recognition at a distance.

This paper presents a novel automatic framework to perform 3D face recognition. The proposed method uses a Simulated Annealing-based approach (SA) for range image registration with the Surface Interpenetration Measure (SIM), as similarity measure, in order to match two face images. The authentication score is obtained by combining the SIM values corresponding to the matching of four different face regions: circular and elliptical areas around the nose, forehead, and the entire face region. Then, a modified SA approach is proposed taking advantage of invariant face regions to better handle facial expressions. Comprehensive experiments were performed on the FRGC v2 database, the largest available database of 3D face images composed of 4,007 images with different facial expressions. The experiments simulated both verification and identification systems and the results compared to those reported by state-of-the-art works. By using all of the images in the database, a verification rate of 96.5 percent was achieved at a False Acceptance Rate (FAR) of 0.1 percent. In the identification scenario, a rank-one accuracy of 98.4 percent was achieved. To the best of our knowledge, this is the highest rank-one score ever achieved for the FRGC v2 database when compared to results published in the literature.

We present a methodology for face segmentation and facial landmark detection in range images. Our goal was to develop an automatic process to be embedded in a face recognition system using only depth information as input. To this end, our segmentation approach combines edge detection, region clustering, and shape analysis to extract the face region, and our landmark detection approach combines surface curvature information and depth relief curves to find the nose and eye landmarks. The experiments were performed using the two available versions of the Face Recognition Grand Challenge database and the BU-3DFE database, in order to validate our proposed methodology and its advantages for 3-D face recognition purposes. We present an analysis regarding the accuracy of our segmentation and landmark detection approaches. Our results were better compared to state-of-the-art works published in the literature. We also performed an evaluation regarding the influence of the segmentation process in our 3-D face recognition system and analyzed the improvements obtained when applying landmark-based techniques to deal with facial expressions.

We present a new 3D reconstruction pipeline for digital preservation of natural and cultural assets. This application requires high quality results, making time and space constraints less important than the achievable precision. Besides the high quality models generated, our work allows an overview of the entire reconstruction process, from range image acquisition to texture generation. Several contributions are shown, which improve the overall quality of the obtained 3D models. We also identify and discuss many practical problems found during the pipeline implementation. Our objective is to help future works of other researchers facing the challenge of creating accurate 3D models of real objects.

Separating data from objects of interest and background is a common procedure in range images applications. Most of the works presented in the literature use image segmentation, either automatic or supervised, to do that. We present a new method to automatically perform this separation without the need of using complex image segmentation techniques. In our approach, we consider that the object is always scanned over a supporting plane. Then, we assume that there is no information below the plane, and the object data is above it. By projecting all points into the supporting plane normal direction, the points on the plane would project at the same value, the points on the object would be spread with values larger than the plane, and there would be very few values below the plane value (due to noise). This allows us to quickly and reliably eliminate the background data from range images.

This paper focuses a comparative evaluation of our framework for 3D face recognition and state-of-theart systems. Our method uses a Simulated Annealing-based approach (SA) for range image registration with the surface interpenetration measure (SIM) as the similarity measure, in order to match two face images. The authentication score is obtained by combining the SIM values corresponding to the matching of four different face regions. Experiments were performed on the FRGC v2 database simulating both verification and identification systems and the obtained results were compared to those reported in the literature. By using all the images in the database, a verification rate of 95.9% was achieved, at a False Acceptance Rate (FAR) of 0.1%. In the identification scenario, a rank-one accuracy of 99.5% was obtained. To our knowledge, this is the best rank-one score obtained on the FRGC v2 database, as compared to previously published results.

The FootScanAge is a non-invasive, more accurate novel approach to automatically estimate the gestational age of newborns. This task is performed by the computational analysis of the features from the newborn plantar surface. The FootScanAge system is composed by two main tools: (1) image processing and (2) data mining. In this paper, we present the image processing tool that extract features of the plantar surface by applying image processing techniques developed to work on the footprint of the newborns, which present some singular characteristics. We included the results that support the determination of an adequate gestational age score by using the data mining tool.

The surface interpenetration measure (SIM) was recently proposed as a promising measure for 3D face matching, although using two limited, small range image databases. In this paper we present novel, more extensive experiments using the SIM in a well-known 3D face database available on the biometric experimentation environment (BEE) to confirm qualitatively that the SIM is a effective, discriminatory measure. The experiments were performed based on range image registration by using two different methods: iterative closest point (ICP) and simulated annealing (SA). By computing the SIM after the registration of two 3D face images one can identify if those images come from the same subject or not. With our SA-based approach we obtained high verification rate scores, which is indeed one of the main goals of the Face Recognition Grand Challenge 2006.

This paper addresses the range image registration problem for views having low overlap and which may include substantial noise. The current state of the art in range image registration is best represented by the well-known iterative closest point (ICP) algorithm and numerous variations on it. Although this method is effective in many domains, it nevertheless suffers from two key limitations: It requires prealignment of the range surfaces to a reasonable starting point and it is not robust to outliers arising either from noise or low surface overlap. This paper proposes a new approach that avoids these problems. To that end, there are two key, novel contributions in this work: a new, hybrid genetic algorithm (GA) technique, including hillclimbing and parallel-migration, combined with a new, robust evaluation metric based on surface interpenetration. Up to now, interpenetration has been evaluated only qualitatively; we define the first quantitative measure for it. Because they search in a space of transformations, GAs are capable of registering surfaces even when there is low overlap between them and without need for prealignment. The novel GA search algorithm we present offers much faster convergence than prior GA methods, while the new robust evaluation metric ensures more precise alignments, even in the presence of significant noise, than mean squared error or other well-known robust cost functions. The paper presents thorough experimental results to show the improvements realized by these two contributions.

This paper presents a novel range image segmentation method employing an improved robust estimator to iteratively detect and extract distinct planar and quadric surfaces. Our robust estimator extends M-estimator Sample Consensus/Random Sample Consensus (MSAC/RANSAC) to use local surface orientation information, enhancing the accuracy of inlier/outlier classification when processing noisy range data describing multiple structures. An efficient approximation to the true geometric distance between a point and a quadric surface also contributes to effectively reject weak surface hypotheses and avoid the extraction of false surface components. Additionally, a genetic algorithm was specifically designed to accelerate the optimization process of surface extraction, while avoiding premature convergence. We present thorough experimental results with quantitative evaluation against ground truth. The segmentation algorithm was applied to three real range image databases and competes favorably against eleven other segmenters using the most popular evaluation framework in the literature. Our approach lends itself naturally to parallel implementation and application in real-time tasks. The method fits well, into several of today's applications in man-made environments, such as target detection and autonomous navigation, for which obstacle detection, but not description or reconstruction, is required. It can also be extended to process point clouds resulting from range image registration.

The paper presents a novel range image segmentation algorithm based on planar surface extraction. The algorithm was applied to common range image databases and was favorably compared against seven other segmentation algorithms using a popular evaluation framework. The experimental results show that, as compared to the other methods, our algorithm presents a good performance in preserving small regions and edge locations when processing noisy images. Our main contribution is an improved robust estimator, derived from the RANSAC and MSAC estimators, whose optimization process is accelerated by a genetic algorithm with a new set of parameters and operations designed to avoid premature convergence.

Most range image registration techniques are based on variants of the ICP (iterative closest point) algorithm. The ICP algorithm has two main drawbacks, the possibility of convergence to a local minimum and the need to prealign the images. Genetic algorithms (GAs) are known to be robust in relation to search and optimization problems and were recently applied to range image registration, providing good convergence results without the constraints observed in the ICP approaches. To improve range image registration by GAs, we explored 3 novel approaches: a hybrid algorithm that combines a GA with hillclimbing heuristics (GH), a parallel migration GA (MGA), and a MGA using hillclimbing (MGH). We also define a new robust evaluation measure, called the surface interpenetration, to compare the obtained registration results. Up to now, interpenetration has been evaluated only qualitatively; we define the first quantitative measure for it. The experimental results show that our methods yield more accurate registration results than either ICP or standard GA approaches.

The determination of gestational age in newborn children is fundamental to the evaluation of their survival possibilities. In many cases, when the qualified pre-natal attendance has not occurred, the post-birth evaluation of the gestational age becomes the only alternative. This paper presents a medical image retrieval system whose goal is to help to confirm a new method (called FootScan) for gestational age determination, through the digital image of the dermatoglyphics from the plantar (sole of the foot) region of the newborn. The system includes a level of knowledge, associated with the database, which stores information from the patients, together with image features and their logical relationships.

We present a range image segmentation algorithm based on a robust estimation technique, the M-estimator sample consensus (MSAC). The algorithm is a parallelizable "global-to-local" approach for the extraction of planar surfaces directly from range images. Solutions to some problems faced when extracting planar surfaces globally are also proposed. Experimental results show the algorithm is robust to image noise in the sense that it is able to preserve object shapes so that neither presmoothing, nor postprocessing steps are required. It also does not rely on MSAC and can be easily adapted to use other robust estimators. Thus, it may be used as a framework to compare robust estimators.

This article presents new improvements to range image segmentation based on edge detection techniques. The developed approach better preserves the object's topology and shape even to noisy images. The algorithm also does not depend on rigid threshold values, thus being useful in unsupervised systems. Experiments were performed in a popular range image database and the results were compared to four other traditional range image segmentation algorithms, demonstrating the efficiency of the proposed algorithm.

A new approach to image segmentation by edge detection is proposed for preserving objects topology and shape while retrieving precisely located, one-pixel-wide edges. The method is based on mean (H) and Gaussian (K) surface curvatures sign maps (HK-sign maps) computed from both registered reflectance and range images, provided by a single sensor. HK-sign maps have been used to identify objects regions on range and intensity images, but not edges, as presented in this work. The combination of the computed range and reflectance edge maps has led to more accurate segmentation results than just by using either of them alone. The proposed algorithm has been tested on real images and compared to four traditional range image segmentation algorithms. Experimental results demonstrate the viability and usefulness of our approach.

Edge detection is an unsolved problem in that, so far, there is no general optimal solution. However, edge detection provides rich information about the scene being observed. This is particularly true in range images, where 3D information is explicit. Many researchers have been taking advantage of edge detection information to improve the segmentation of range images by integrating edge detection with other different segmentation techniques. This paper presents a methodology to perform edge detection in range images in order to provide a reliable and meaningful edge map, which helps to guide and improve range image segmentation by clustering techniques. The obtained edge map leads to three important improvements: (1) the definition of the ideal number of regions to initialize the clustering algorithm; (2) the selection of suitable initial cluster centers; and (3) the successful identification of distinct regions with similar features. Experimental results that substantiate the effectiveness of this work are presented.