Aline Deruyver

Constraint satisfaction problem with bilevel constraint: application to interpretation of over-segmented images

Abstract In classical finite-domain constraint satisfaction problems, the assumption made is that only one value is associated with only one variable. For example, in pattern recognition one variable is associated with only one segmented region. However, in practice, regions are often oversegmented which results in failure of any one to one mapping. This paper proposes a definition of finite-domain constraint satisfaction problems with bilevel constraints in order to take into account a many to one relation between the values and the variables. The additional level of constraint concerns the data assigned to the same complex variable. Then, we give a definition of the arc-consistency problem for bilevel constraint satisfaction checking. A new algorithm for arc consistency to deal with these problems is presented as well. This extension of the arc-consistency algorithm retains its good properties and has a time complexity in O( en 3 d 2 ) in the worst case. This algorithm was tested on medical images. These tests demonstrate its reliability in correctly identifying the segmented regions even when the image is over-segmented.

Adaptive pyramid and semantic graph: knowledge driven segmentation

A method allowing to integrate syntactic and semantic approaches in an automatic segmentation process is described. This integration is possible thanks to the formalism of graphs. The proposed method checks the relevancy of merging criteria used in an adaptive pyramid by matching the obtained segmentation with a semantic graph describing the objects that we look for. This matching is performed by checking the arc-consistency with bilevel constraints of the chosen semantic graph. The validity of this approach is experimented on synthetic and real images.

Near-field measurements of vegetation by laser-induced fluorescence imaging

In this paper, a validation of a new UV-A laser-induced fluorescence imaging system implemented in an all-road car for near-field remote sensing of vegetation will be presented. It has been developed as a part of a European Community Program INTERREG II and is consisting of three main parts: excitation, detection and control units. The excitation source is a frequency tripled Nd:YAG laser and the laser spot size is adjusted via a variable beam expander. Fluorescence images are recorded at four characteristic fluorescence bands: 440, 520, 690 and 740 nm with a gated intensified digital CCD camera. The laser head and camera are situated on a directed in site and azimuth platform which can be high up to 6 meters. The platform positioning, localization and distance detection, spot size determination and adjustment, focus, sharpness, selection of the filter, laser and camera synchronization, gain of the intensifier, real time visualization of images, acquisition time are controlled by a newly developed software which allows also image storage, analysis and treatment. Examples of remote sensing fluorescence images from several plant species recorded at a distance of 10 - 30 m will be given and discussed further in this paper.

Template-Based Hierarchical Building Extraction

Automatic building extraction is an important field of research in remote sensing. This letter introduces a new object-based building extraction approach. So far, many object-based algorithms for building extraction have been proposed. However, these algorithms mainly operate in two phases: object construction and building extraction. The majority of these algorithms heavily relies on the object construction process, mainly due to the lack of interaction between the two steps. To overcome these drawbacks, we introduce a new hierarchical approach based on building templates. Carried out experiments on data sets of images from the urban area of Strasbourg show the benefits of our approach.

Hierarchical classification-based region growing (HCBRG): a collaborative approach for object segmentation and classification

Object-based image classification approaches heavily rely on the segmentation process. However, the lack of interaction between both segmentation and classification steps is one of the major limits of these approaches. In this paper, we introduce a hierarchical classification based on a region growing approach driven by expert knowledge represented in a concept hierarchy. In order to overcome the region growings limits, a first classification will associate a confidence score to each region in the image. This score will be used through an iterative step, which allows interaction between segmentation and classification at each iteration. Carried out experiments on a Quickbird image show the benefits of the introduced approach.

Hypersonic shock waves and hybridization of a-C:H thin films

A distributed electron cyclotron resonance plasma reactor powered by a microwave generator operating at 2.45 GHz was used to deposit a-C:H films at room temperature on rf biased 〈100〉 Si substrates. Modifying substrate bias, substrate current density and composition of the precursor gas enabled the average deposited energy density to be varied. The physical properties of a-C:H were investigated using atomic force microscopy (AFM), x-ray photoelectron spectroscopy, and electron energy loss spectroscopy (EELS). The experimental results were correlated with the predictions of the binary collision theory. The influence of the deposited energy density on the nucleation and growth processes was investigated using both pure C2H2 and C2H2 mixed with Ar. The sp3 nucleation process is shown to be stimulated by high energy density cascades generated by (C2H2)+ and Ar+ ions. For the pure acetylene plasma, the AFM topography displays a random network of circular, crater-like objects close to 1 μm in diameter. These ob...

Watershed and adaptive pyramid for determining the apple's maturity state

A new method of image processing, developed within the framework of the automatic evaluation of the state of maturity of fruit, is presented. It has been shown that maturity spots can be seen on images of fruit obtained by laser-induced fluorescence. The presented method is able to detect and to quantify these spots automatically. The initial image processing is provided by a watershed algorithm. In order to refine the obtained segmentation, a multiresolution method is developed. It is based on the representation of data with an adaptive pyramid of region adjacency graphs (RAG) associated with a multi-criteria approach. This approach allows regions to be merged and only the information corresponding to the maturity spot to be stored. Interesting experimental results have been obtained on a set of 20 images of apples.

A system to detect residential area in multispectral satellite images

In this paper, we propose a new solution to extract complex structures from High-Resolution (HR) remote-sensing images. We propose to represent shapes and there relations by using region adjacency graphs. They are generated automatically from the segmented images. Thus, the nodes of the graph represent shape like houses, streets or trees, while arcs describe the adjacency relation between them. In order to be invariant to transformations such as rotation and scaling, the extraction of objects of interest is done by combining two techniques: one based on roof color to detect the bounding boxes of houses, and one based on mathematical morphology notions to detect streets. To recognize residential areas, a model described by a regular language is built. The detection is achieved by looking for a path in the region adjacency graph, which can be recognized as a word belonging to the description language. Our algorithm was tested with success on images from Quickbird satellite representing the urban area of Strasbourg (France) at different spatial resolution.

A morphological model for automatic edge detection: comparison with ad-hoc hysteresis thresholding

In edge detection, the application of gradient operators and other derived operators is commonly used. In all cases, the result is a grey level image which needs to be binarised. In this paper, an automatic method of binarisation based on a morphological model of edges is proposed. To prove the robustness of this approach, a testing method has been developed. This method compares the performance of the new approach with two manual methods often used: the application of an optimal threshold value, and the application of the hysteresis method. The comparison is made very accurately by computing statistical criteria such as specificity, sensitivity, predictivity and the percentage of well classified edges. The set of test images is made up of synthetic images with different kinds of noise (uniform or Gaussian) and an increasing amount of edges. This approach was applied on medical images and common images as well. The results demonstrate its reliability.

A segmentation technique for cerebral NMR images

Presents a method of automatic segmentation of cerebral anatomic regions in NMR images. NMR image segmentation is of great interest. The morphological studies are easier and anomalies can be detected as diagnostic indicators of disease. The automatic approach described here works without a priori knowledge. It increases the contrast between grey and white matter by using a local histogram equalization. Then it uses morphological operators to remove the noise from the image and then to refine the detected anatomic regions. This method has been applied to more than 40 images. Each time, the result was as good as that obtained with manual segmentation.<<ETX>>