Felipe Calderero

Region Merging Techniques Using Information Theory Statistical Measures

The purpose of the current work is to propose, under a statistical framework, a family of unsupervised region merging techniques providing a set of the most relevant region-based explanations of an image at different levels of analysis. These techniques are characterized by general and nonparametric region models, with neither color nor texture homogeneity assumptions, and a set of innovative merging criteria, based on information theory statistical measures. The scale consistency of the partitions is assured through i) a size regularization term into the merging criteria and a classical merging order, or ii) using a novel scale-based merging order to avoid the region size homogeneity imposed by the use of a size regularization term. Moreover, a partition significance index is defined to automatically determine the subset of most representative partitions from the created hierarchy. Most significant automatically extracted partitions show the ability to represent the semantic content of the image from a human point of view. Finally, a complete and exhaustive evaluation of the proposed techniques is performed, using not only different databases for the two main addressed problems (object-oriented segmentation of generic images and texture image segmentation), but also specific evaluation features in each case: under- and oversegmentation error, and a large set of region-based, pixel-based and error consistency indicators, respectively. Results are promising, outperforming in most indicators both object-oriented and texture state-of-the-art segmentation techniques.

General region merging approaches based on information theory statistical measures

This work presents a new statistical approach to region merging where regions are modeled as arbitrary discrete distributions, directly estimated from the pixel values. Under this framework, two region merging criteria are obtained from two different perspectives, leading to information theory statistical measures: the Kullback-Leibler divergence and the Bhattacharyya coefficient. The developed methods are size-dependent, which assures the size consistency of the partitions but reduces their size resolution. Thus, a size-independent extension of the previous methods, combined with a modified merging order, is also proposed. Additionally, an automatic criterion to select the most statistically significant partitions from the whole merging sequence is presented. Finally, all methods are evaluated and compared with other state-of-the-art region merging techniques.

Recovering Relative Depth from Low-Level Features Without Explicit T-junction Detection and Interpretation

This work presents a novel computational model for relative depth order estimation from a single image based on low-level local features that encode perceptual depth cues such as convexity/concavity, inclusion, and T-junctions in a quantitative manner, considering information at different scales. These multi-scale features are based on a measure of how likely is a pixel to belong simultaneously to different objects (interpreted as connected components of level sets) and, hence, to be occluded in some of them, providing a hint on the local depth order relationships. They are directly computed on the discrete image data in an efficient manner, without requiring the detection and interpretation of edges or junctions. Its behavior is clarified and illustrated for some simple images. Then the recovery of the relative depth order on the image is achieved by global integration of these local features applying a non-linear diffusion filtering of bilateral type. The validity of the proposed features and the integration approach is demonstrated by experiments on real images and comparison with state-of-the-art monocular depth estimation techniques.

Multiview depth coding based on combined color/depth segmentation

In this paper, a new coding method for multiview depth video is presented. Considering the smooth structure and sharp edges of depth maps, a segmentation based approach is proposed. This allows further preserving the depth contours thus introducing fewer artifacts in the depth perception of the video. To reduce the cost associated with partition coding, an approximation of the depth partition is built using the decoded color view segmentation. This approximation is refined by sending some complementary information about the relevant differences between color and depth partitions. For coding the depth content of each region, a decomposition into orthogonal basis is used in this paper although similar decompositions may be also employed. Experimental results show that the proposed segmentation based depth coding method outperforms H.264/AVC and H.264/MVC by more than 2dB at similar bitrates.

Hierarchical fusion of color and depth information at partition level by cooperative region merging

A high level scheme for information fusion to create hierarchical region-based image representations based on a region merging process is presented. The strategy is based on an iterative evolution where the different merging criteria work independently and cooperate at the partition level to obtain a further consensus that increases the reliability of the resulting partitions. This cooperative scheme is applied to the creation of hierarchical region-based representations of the image based on color and depth information. The proposed technique is compared with approaches using only one source of information or linear combinations of both, in datasets with ground truth as well as estimated disparity information.

Hierarchical segmentation of vegetation areas in high spatial resolution images by fusion of multispectral information

A new region-based methodology for the automated extraction and hierarchical segmentation of vegetation areas into high spatial resolution images is proposed. This approach is based on the iterative and cooperative fusion of the independent segmentation results of equal or different resolution spectral bands, combined with an unsupervised classification into vegetation and no-vegetation regions. The result is a hierarchy of partitions with most relevant information at different levels of resolution of the vegetation areas. In addition, the high flexibility of the scheme allows different configurations depending on the final purpose. For instance, considering the size of the vegetation areas into the hierarchy, or prioritizing the information into the high resolution panchromatic band to improve the accuracy of both vegetation extraction and segmentation. This general tool for vegetation analysis is tested into high spatial resolution images from IKONOS and QuickBird satellites.

Multiscale Analysis of Similarities between Images on Riemannian Manifolds

In this paper we study the problem of comparing two patches of an image defined on a Riemannian manifold, which can be defined by the image domain with a suitable metric depending on the image. The size of the patch will not be determined a priori, and we identify it with a variable scale. Our approach can be considered as a nonlocal extension (comparing two points) of the multiscale analyses defined using the axiomatic approach by Alvarez et al. [Arch. Ration. Mech. Anal., 123 (1993), pp. 199--257]. Following this axiomatic approach, we can define a set of similarity measures that appear as solutions of a degenerate partial differential equation. This equation can be further specified in the linear case, and we observe that it contains as a particular instance the case of using weighted Euclidean distances as comparison measures. Finally, we discuss the case of some morphological scale spaces that exhibit a higher complexity.

Multispectral Cooperative Partition Sequence Fusion for Joint Classification and Hierarchical Segmentation

In this letter, a region-based fusion methodology is presented for joint classification and hierarchical segmentation of specific ground cover classes from high-spatial-resolution remote sensing images. Multispectral information is fused at the partition level using nonlinear techniques, which allows the different relevance of the various bands to be fully exploited. A hierarchical segmentation is performed for each individual band, and the ensuing segmentation results are fused in an iterative and cooperative way. At each iteration, a consensus partition is obtained based on information theory and is combined with a specific ground cover classification. Here, the proposed approach is applied to the extraction and segmentation of vegetation areas. The result is a hierarchy of partitions with the most relevant information of the vegetation areas at different levels of resolution. This system has been tested for vegetation analysis in high-spatial-resolution images from the QuickBird and GeoEye satellites.

A method to reconstruct activation wavefronts without isotropy assumptions using a level sets approach

We report on an investigation into using a Level Sets based method to reconstruct activation wavefronts at each time instant from measured potentials on the body surface. The potential map on the epicardium is approximated by a two level image and the inverse problem is solved by evolving a boundary, starting from an initial region, such that a filtered residual error is minimized. The advantage of this method over standard activation-based solutions is that no isotropy assumptions are required. We discuss modifications of the Level Sets method used to improve accuracy, and show the promise of this method via simulation results using recorded canine epicardial data.

Region merging parameter dependency as information diversity to create sparse hierarchies of partitions

Region merging techniques usually include parameters that may be used to optimize or adapt the algorithm to a specific image type. Although, an appropriate tuning may provide a significant improvement, it also introduces a severe performance dependency on the parameter setting. The goal of this work is to transform the parameter dependency into an increase of accuracy and stability of the segmentation results. The idea is to use different parameter settings as specific type of diversity in an information fusion process based on a cooperative region merging approach. The potential of this parameter removal strategy is objectively evaluated on a set of state-of-the-art information theoretical region merging techniques for the removal of parameters: (i) in the region model, and (ii) in the merging order.