Serge Beucher

WATERSHED, HIERARCHICAL SEGMENTATION AND WATERFALL ALGORITHM

A major drawback when using the watershed transformation as a segmentation tool comes from the over-segmentation of the image. Over-segmentation is produced by the great number of minima embedded in the image or in its gradient. A powerful technique has been designed to suppress over-segmentation by a primary selection of markers pointing out the regions or objects to be segmented in the image. However, this approach can be used only if we are able to compute the marker set before applying the watershed transformation. But, in many cases and especially for complex scenes, this is not possible and an alternative technique must be used to reduce the over-segmentation. This technique is based on mosaic images and on the computation of a watershed transform on a valued graph derived from the mosaic images. This approach leads to a hierarchical segmentation of the image and considerably reduces over-segmentation.

Road segmentation and obstacle detection by a fast watershed transformation

This paper presents the work, performed at CMM as part of the European PROMETHEUS project, concerning road/lane segmentation and obstacle detection in a dynamic scene. Road/lane segmentation is used to select ROI where obstacle detection algorithms are applied. This segmentation is made up of a temporal filter, an edge detector and a watershed transformation. It produces a marker of the current traffic lane. This marker is then used for calculating a road model which is continuously updated. The obstacle detection uses various criteria: existence of a darker region marking a vehicle, size and contrast of obstacles and symmetry. These two processes use the watershed transformation which is usually the slowest part of the segmentation. To speed it up, a very efficient enhancement of the algorithm, based on anamorphosis, is presented.

Numerical residues

Binary morphological transformations based on the residues (ultimate erosion, skeleton by openings, etc.) and their associated functions which are based on the analysis of the residue evolution in every point of the image are extended to functions. In this approach, the associated function indicates the value of the residue index for which the evolution is the most important. The approach also has the advantage of supplying effective tools for shape analysis and of allowing the definition of new residual transforms together with their associated functions. Two of these numerical residues will be introduced, called, respectively, ultimate opening and quasi-distance and, through some applications, the interest and efficiency of these operators will be illustrated. Finally, this residual approach will be extended to more complex operators.

Road tracking, lane segmentation and obstacle recognition by mathematical morphology

Presents the work performed at the CMM as part of the European PROMETHEUS project. The research deals with the road edge detection, lane segmentation and obstacle recognition in a dynamic scene acquired by a monochrome monocular camera. The image processing is based on morphological segmentation tools. The experiments on over a thousand images show that the approach works well on difficult cases such as dense traffic, and roads without land markers.<<ETX>>

FAST IMPLEMENTATION OF WATERFALL BASED ON GRAPHS

The waterfall algorithm is a contrast-based hierarchical segmentation approach. In this paper we propose an efficient implementation based on the minimum spanning tree of the neighborhood graph. Furthermore, other hierarchies are proposed and compared to the original version of the algorithm.

Marker-controlled segmentation: an application to electrical borehole imaging

An application of marker-controlled segmentation in petroleum engineering is presented. The images to be segmented originate from high-resolution conductivity measurements of borehole walls. These measurements reflect the composition and structure of the rock formation through which the well was drilled. In this application, we detect and measure small cavities, or vugs, in the walls. We use the tools provided by mathematical morphology. Our strategy is based on gradient image modification using markers and on the watershed transformation. First, the vugs are automatically marked, as well as the background. These markers together delineate areas of interest in which we know there is one contour per vug. To find the vug contour and perform measurements, we modify the gradient image in such a way that only a single edge is kept between the vug and the background markers. We perform the final step of edge detection using the watershed transformation of the modified gradient image. The final result is one closed contour per marked vug. This strategy is presented in detail, experimental results are shown, and artifact elimination is discussed.

Digital skeletons in Euclidean and geodesic spaces

Abstract This paper is an attempt to bridge the gap between two classes of skeleton algorithms widely used in mathematical morphology: the skeleton by maximal balls and the skeleton by thinnings. First, we show that the first type of skeleton, usually resulting from morphological openings according to Lantuejouls formula, can also be obtained by non-homotopic thinnings. Moreover, these thinnings are not sequential but based on the intersection of elementary thinnings. The second step is a restrictive selection among the previous structuring elements which preserve homotopy. It leads to the definition of a connected skeleton containing the skeleton by maximal balls. This algorithm is given in the 2D Euclidean space for both the hexagonal and square grids. This skeleton combines the advantages of the two classes of skeletons and avoids the main drawbacks involved by rotating thinnings for building connected skeletons. The extension of this definition to geodesic spaces is discussed in the third part. The geodesic thinnings are introduced to define a connected skeleton containing the skeleton by maximal balls.

An Efficient Hardware Architecture without Line Memories for Morphological Image Processing

In this paper, we present a novel hardware architecture to achieve erosion and dilation with a large structuring element. We are proposing a modification of HGW algorithm with a block mirroring scheme to ease the propagation and memory access and to minimize memory consumption. It allows to suppress the needs for backward scanning and gives the possibility for hardware architecture to process very large lines with a low latency. It compares well with the Lemonniers architecture in terms of ASIC gates area and shows the interest of our solution by dividing the circuit area by an average of 10.

Morphological tools for indexing video documents

This paper proposes a local algorithm using morphological operators, which leads to several useful tools for indexing video documents. It consists of a local computation of a similarity criterion between two successive frames of a sequence, followed by the study of the temporal evolution curve of this criterion for the whole sequence. From this curve, shot transitions are extracted by means of a powerful morphological filter, the inf top-hat. At this point, we have built a cut detection tool. The local computation together with strong morphological filtering leads to a very simple, fast and computationally efficient algorithm, with a high detection rate in the case of cuts for a small level of false detections. This algorithm also gives access to a spatial model of the transition and to a selection of key frames for each shot. By applying the local similarity measure to the key frames, two other tools are built to detect inner shot changes and syntactically related shots. Finally, the relation detection is used in a newscaster detection tool.

Marker-controlled picture segmentation applied to electrical logging images

This paper presents an application of marker-controlled segmentation in petroleum engineering. The images to be segmented originate from high resolution conductivity measurements of borehole walls. These measurements reflect the composition and structure of the rock formation through which the well was drilled. In this application, we detect and measure small cavities in the walls. These cavities are called vugs. We use the tools provided by mathematical morphology. Our strategy is based on gradient image modification using markers and on the watershed transformation. First, the vugs are automatically marked, as well as the background. These markers together delineate areas of interest in which we know there is one contour per vug. In order to find the vug contour and perform measurements, we modify the gradient image in such a way that only a single edge is kept between the vug and the background markers. We perform the final step of edge detection using the watershed transformation of the modified gradient image. The final result is one closed contour per marked vug. We present this strategy in detail, show experimental results and discuss artifact elimination.