Kamal Nasreddine

Variational shape matching for shape classification and retrieval

In this paper we define a multi-scale distance between shapes based on geodesics in the shape space. The proposed distance, robust to outliers, uses shape matching to compare shapes locally. The multi-scale analysis is introduced in order to address local and global variabilities. The resulting similarity measure is invariant to translation, rotation and scaling independently of constraints or landmarks, but constraints can be added to the approach formulation when needed. An evaluation of the proposed approach is reported for shape classification and shape retrieval on the part B of the MPEG-7 shape database. The proposed approach is shown to significantly outperform previous works and reaches 89.05% of retrieval accuracy and 98.86% of correct classification rate.

An optimal elastic partial shape matching via shape geodesics

This paper presents a planar curve matching framework based on computing similarities between shape parts. We propose an elastic similarity measure issued from shape geodesics in the shape space. As the transition from global matching to partial matching leads to additional difficulties, we bypass them with a shape decomposition process based on the discrete curve evolution (DCE). This decomposition aims to obtain significant parts to match and it leads to a robust and efficient 2D shape matching algorithm. The comparison of the proposed method with the state of the art demonstrates its ability to handle elastic deformations leading to an overall optimal partial correspondence between shapes.

Variational 1D signal registration and shape geodesics for shape classification: Application to marine biological archives

When two 1D signals are compared, they must be represented in the same reference system. In most cases, biological signals present a big interindividual variability that should be eliminated in order to compare them properly. This variability can be erased by aligning the signals. A robust variational setting is proposed for 1D signal registration and applied to the computation of shape geodesics for shape classification issues. For validation purposes, experiments are carried out on real signals and shapes issued from marine biological archives which depict a high interindividual variability such that registration-based approaches are of particular interest.

Geodesics-Based Image Registration: Applications To Biological And Medical Images Depicting Concentric Ring Patterns

In many biological or medical applications, images that contain sequences of shapes are common. The existence of high inter-individual variability makes their interpretation complex. In this paper, we address the computer-assisted interpretation of such images and we investigate how we can remove or reduce these image variabilities. The proposed approach relies on the development of an efficient image registration technique. We first show the inadequacy of state-of-the-art intensity-based and feature-based registration techniques for the considered image datasets. Then, we propose a robust variational method which benefits from the geometrical information present in this type of images. In the proposed non-rigid geodesics-based registration, the successive shapes are represented by a level-set representation, which we rely on to carry out the registration. The successive level sets are regarded as elements in a shape space and the corresponding matching is that of the optimal geodesic path. The proposed registration scheme is tested on synthetic and real images. The comparison against results of state-of-the-art methods proves the relevance of the proposed method for this type of images.

Shape geodesics for boundary-based object recognition and retrieval

In this paper we define a distance between shapes based on geodesics in shape space. The proposed distance, robust to outliers, uses shape matching to compare shapes locally. Multiscale analysis is introduced in order to avoid problems of local and global variabilities. The resulting similarity measure is invariant to translation, rotation and scaling independently of constraints or landmarks, but constraints can be added to the approach formulation when needed. An evaluation of the proposed approach is reported for shape classification and retrieval on a complex benchmark shape database. It demonstrates in both cases that previous work is outperformed.

Geodesics-based statistical shape analysis

In this paper, we describe a statistical shape analysis founded on a robust elastic metric. The proposed metric is based on geodesics in the shape space. Using this distance, we formulate a variational setting to estimate intrinsic mean shape viewed as the perfect pattern to represent a set of given shapes. By applying a geodesic-based shape warping, we generate a principal component analysis (PCA) able to capture nonlinear shape variability. Indeed, the proposed approach better reflects the main modes of variability of the data. Therefore, characterizing dominant modes of individual shape variations is conducted well through the reconstruction process. We demonstrate the efficiency of our approach with an application on a GESTURES database.

A comparative evaluation of segmentation methods for dendritic cells identification from microscopic images

Public health is one of the major concerns at the world level. Toxicology is an extremely challenging issue regarding that toxic substances are harmful to human health. In fact, toxicology studies are indispensable to evaluate the toxic effects on humans. Currently, a new evaluation technique based on the analysis of dendritic cells in vitro has been found by researchers. This analysis that remains purely visual is a tedious process, subjective and time-consuming. Therefore, an assessment tool for the analysis of toxic impact using automatic processing techniques by image analysis can be greatly useful for expert biologists. The foremost aim of this paper is to propose two segmentation approaches of dendritic cells from microscopic images and to present a comparative evaluation of them. The first suggested algorithm is based on automatic thresholding and mathematical morphology, while the second one combines the k-means clustering, thresholding and mathematical morphology based operations. For validation purposes, four performance measures were used to assess the obtained segmentation results with the ground truth images, elaborated by expert. Quantitatively, results show that the two suggested algorithms are efficient in identifying dendritic cells from 26 gray-scale images with a segmentation accuracy of 99.00 % and 99.37%, respectively.

Non-rigid registration of shape sequence images: Applications to biological and medical images

This paper deals with the automation of the analysis of images depicting shape sequences. Here, a robust method for matching shape sequence images is developed. First, the successive shapes are represented by a level-set representation, then the algorithm of registration is carried out on this level-set representation. It considers the levels as elements in a shape space and the corresponding matching is that of the optimal geodesic path. This algorithm of registration is tested on synthetic images and real images issued from biological and medical applications where the intensity-based registration fails.

Segmentation of dendritic cells from microscopic images using mathematical morphology

For many years, biologists have been interested in toxicology to assess the effects of contamination on humans. In recent years, researchers have found a new evaluation technique based on the analysis of dendritic cells in vitro. Up to now the analysis conducted on these cells remains purely visual in nature. Therefore, it is subjective and time-consuming because of the different morphological features of the cells. Here, we suggest to use automatic processing techniques by image analysis. The foremost goal of this paper is to suggest an assessment tool for the analysis of immunotoxic effects of food contaminants (mycotoxins) on the immune system using automatic segmentation techniques of microscopic images of dendritic cells. The suggested method is based on automatic thresholding and mathematical morphology. For validation purposes, an experimental study is carried out on 55 microscopic images of dendritic cells visually analyzed by an expert in order to make comparisons or to have a reference segmentation of the cells. Results show that the proposed approach is efficient in identifying dendritic cells with a segmentation accuracy of 95%.

Robust statistical shape analysis based on the tangent shape space

In this paper, we propose and develop a robust formulation for statistical shape analysis based on an elastic metric between closed planar curves. The proposed solution is founded on robustifying the inverse exponential map that links the preshape space to the tangent shape space relatively to a reference point. Applying this robust transition map, we obtain a rectified version of the shape database cleaned from aberrant points and more adequate for statistical analysis. Hence, we derive a new tangent PCA which we call a Robust Tangent PCA (RTPCA) where the main modes reflect the variability of the data with a resistance to outliers that may affect a classical analysis. We illustrate the capability of our approach with an application on the Kimia-HAND database.