Detecting Corners for 3D Objects

In this paper, a corner detection algorithm for 3D objects is presented. This algorithm is an extension of corner detection scheme for planar objects (Chetrikov & Zsabo, 1999). This algorithm finds corners and other high curvature points for 3D objects.

Edge Detection by Maximum Entropy

In the edge detection, the classical operators based on the derivation are sensitive to noise which causes detection errors. It is even more erroneous in the case of omnidirectional images, due to geometric distortions caused by the used sensors. This paper proposes a statistical method of edge detection invariant to image resolution applied to omnidirectional images without preliminary treatments. It is based on the entropy measure. The authors compared its behavior with existing methods on omnidirectional images and perspectives images. The criteria of comparisons are the parameters of Fram and Deutsch. For omnidirectional images, the authors used two types of neighborhood: fixed and adapted to the parameters of the sensor. The authors compared the results of detection visually. The tests are performed on grayscale images.

Automatic Facial Expression Recognition by Facial Parts Location with Boosted-LBP

This paper proposes an automatic facial expression recognition system, which uses new methods in both face detection and feature extraction. In this system, considering that facial expressions are related to a small set of muscles and limited ranges of motions, the facial expressions are recognized by these changes in video sequences. First, the differences between neutral and emotional states are detected. Faces can be automatically located from changing facial organs. Then, LBP features are applied and AdaBoost is used to find the most important features for each expression on essential facial parts. At last, SVM with polynomial kernel is used to classify expressions. The method is evaluated on JAFFE and MMI databases. The performances are better than other automatic or manual annotated systems.

An Evaluation Framework and a Benchmark for Multi/Hyperspectral Image Compression

This paper benchmarks three multi/hyperspectral image compression approaches: the classic Multi-2D compression approach and two different implementations of 3D approach (Full 3D and Hybrid). All approaches are combined with a spectral PCA decorrelation stage to optimize performance. These three compression approaches are compared within a larger comparison framework than the conventionally used PSNR, which includes eight metrics divided into three families. The comparison is carried out with regard to variations in bitrates, spatial, and spectral dimensions variations of images. The time and memory consumption difference between the three approaches is also discussed. Results of this comparison show the weaknesses and strengths of each approach.

Construction of 3D Triangles on Dupin Cyclides

This paper considers the conversion of the parametric Bézier surfaces, classically used in CAD-CAM, into patched of a class of non-spherical degree 4 algebraic surfaces called Dupin cyclides, and the definition of 3D triangle with circular edges on Dupin cyclides. Dupin cyclides was discovered by the French mathematician Pierre-Charles Dupin at the beginning of the 19th century. A Dupin cyclide has one parametric equation, two implicit equations, and a set of circular lines of curvature. The authors use the properties of these surfaces to prove that three families of circles (meridian arcs, parallel arcs, and Villarceau circles) can be computed on every Dupin cyclide. A geometric algorithm to compute these circles so that they define the edges of a 3D triangle on the Dupin cyclide is presented. Examples of conversions and 3D triangles are also presented to illustrate the proposed algorithms.

A Robust Embedding Scheme and an Efficient Evaluation Protocol for 3D Meshes Watermarking

This paper proposes two main contributions. In the first one, a 3D mesh watermarking using Maximally Stable Meshes detection and multi-signatures embedding is presented. The originality of this scheme is to detect the attack type applied on marked mesh. In plus, it is robust against numerous attacks, blind and invisible. The proposed scheme uses the Maximally Stable meshes (MSMs) to insert signature. After MSMs detection using an extension of Maximally Stable Efficient Regions, three MSMs are selected to be marked. Then, three different signatures are embedded using three different watermarking schemes. This embedding allows knowing the type of the applied attack by detecting which of the signatures resisted. In more, it maximizes robustness by profiting from advantages of every scheme. The second contribution is a new evaluation protocol for 3D watermarking which allows generating a performance score for 3D mesh watermarking schemes. This protocol is based on six criteria having different weights in performance score computing. Finally, this protocol is used to evaluate the proposed watermarking scheme and to compare it with other algorithms. The obtained results verified the good performances of the proposed algorithm which presents the highest score.

Evaluation Approach of Arabic Character Recognition

This paper proposes and contributes towards designing a complete system for off-line Arabic character recognition. The proposed system is specifically meant for Arabic handwriting recognition, but it equally works for the typed character recognition. It has various phases including preprocessing and segmentation. It also includes thinning phase and finds vertical and horizontal projection profiles. The recognition phase is managed by genetic algorithm. The genetic algorithm stands on feature extraction algorithm that defines six features for each segment. The algorithm, for Arabic handwriting recognition, obtained 90.46 recognition rate. The proposed system has been compared with other systems in the literature. It has achieved the second best recognition rate.

Direct 3D Information Determination in an Uncalibrated Stereovision System by Using Evolutionary Algorithms

This paper proposes a 3D panoramic shape reconstruction method based on an uncalibrated stereovision system (USS) composed of five cameras circularly located around the object to be analysed. First, some interesting points are detected from markers placed on the object such that they are visible by two successive cameras of the USS. These points are then matched on both images acquired by a couple of successive cameras. This process is repeated for all the couples of cameras. Second, by using an evolutionary algorithm, the depth values of the different interesting points are calculated. A comparison with a traditional method based on calibrated cameras validates the accuracy of 3D information provided by the proposed method. Finally, by combining all the interesting points, a panoramic view of the object is obtained.

Adapted Approach for Omnidirectional Egomotion Estimation

Egomotion estimation is based principally on the estimation of the optical flow in the image. Recent research has shown that the use of omnidirectional systems with large fields of view allow overcoming the limitation presented in planar-projection imagery in order to address the problem of motion analysis. For omnidirectional images, the 2D motion is often estimated using methods developed for perspective images. This paper adapts motion field calculated using adapted method which takes into account the distortions existing in the omnidirectional image. This 2D motion field is then used as input to the egomotion estimation process using spherical representation of the motion equation. Experimental results are shown and comparison of error measures are given to confirm that succeeded estimation of camera motion will be obtained when using an adapted method to estimate optical flow.

A Semi-Supervised Metric Learning for Content-Based Image Retrieval

In this paper, the authors propose a kernel-based approach to improve the retrieval performances of CBIR systems by learning a distance metric based on class probability distributions. Unlike other metric learning methods which are based on local or global constraints, the proposed method learns for each class a nonlinear kernel which transforms the original feature space to a more effective one. The distances between query and database images are then measured in the new space. Experimental results show that the kernel-based approach not only improves the retrieval performances of kernel distance without learning, but also outperforms other kernel metric learning methods.