Frederic Truchetet

Rational multiresolution analysis and fast wavelet transform: application to wavelet shrinkage denoising

Abstract This paper presents a contribution to rational multiresolution analysis (MRA). The rational analysis allows a better adaptation of scale factors to signal components than the dyadic one. The theory of rational MRA is reviewed and a pyramidal algorithm for fast rational orthogonal wavelet transform is proposed. Both, the analysis and synthesis parts of the process are detailed. Examples of scaling and wavelet functions and associated filters are given. Moreover, dealing with filters defined in Fourier domain, the implementation of the algorithm in this domain is described. Then, the study is extended to the 2D separable case in order to give a more conclusive presentation of the rational MRA. In order to illustrate the potential of rational analysis for signal and image processing, some results given by wavelet shrinkage denoising based on the ‘SURE’ thresholding method are presented.

MESH COMPARISON USING ATTRIBUTE DEVIATION METRIC

We propose a mesh comparison method using a new attribute deviation metric. The considered meshes contain geometrical and appearance attributes (material color, texture, temperature, etc.). The proposed deviation metric computes local differences between the attributes of two meshes. A mesh comparison assessment can be done easily and quickly using this metric. The techniques proposed are applicable in a number of ways, e.g. 3D matching and registration, and the example described in the paper is the simplification of a surface by iteratively reducing its complexity according to an error metric. The results are presented showing the success of the algorithm through comparisons with other measures and with three different simplification algorithms.

Scanning from heating: 3D shape estimation of transparent objects from local surface heating

Today, with quality becoming increasingly important, each product requires three-dimensional in-line quality control. On the other hand, the 3D reconstruction of transparent objects is a very difficult problem in computer vision due to transparency and specularity of the surface. This paper proposes a new method, called Scanning From Heating (SFH), to determine the surface shape of transparent objects using laser surface heating and thermal imaging. Furthermore, the application to transparent glass is discussed and results on different surface shapes are presented.

A Nonlinear Derivative Scheme Applied to Edge Detection

This paper presents a nonlinear derivative approach to addressing the problem of discrete edge detection. This edge detection scheme is based on the nonlinear combination of two polarized derivatives. Its main property is a favorable signal-to-noise ratio (SNR) at a very low computation cost and without any regularization. A 2D extension of the method is presented and the benefits of the 2D localization are discussed. The performance of the localization and SNR are compared to that obtained using classical edge detection schemes. Tests of the regularized versions and a theoretical estimation of the SNR improvement complete this work.

Generalization of Canny---Deriche filter for detection of noisy exponential edge

This paper presents a generalization of the Canny-Deriche filter for ramp edge detection with optimization criteria used by Canny (signal-to-noise ratio, localization, and suppression of false responses). Using techniques similar to those developed by Deriche, we derive a filter which maximizes the product of the first two criteria under the constraint of the last one. The result is an infinite length impulse response filter which leads to a stable third-order recursive implementation. Its performance shows an increase of the signal-to-noise ratio in the case of blurred and noisy images, compared to the results obtained from Deriches filter.

Review of industrial applications of wavelet and multiresolution-based signal and image processing

Twenty five years after the seminal work of Jean Morlet, the wavelet transform, multiresolution analysis, and other space-frequency or space-scale approaches are considered standard tools by researchers in image processing. Many applications that point out the interest of these techniques have been proposed. We review the recent published work dealing with industrial applications of the wavelet and, more generally speaking, multiresolution analysis. We present more than 190 recent papers.

Modelling agronomic images for weed detection and comparison of crop/weed discrimination algorithm performance

A new method for weed detection based on modelling agronomic images taken from a virtual camera placed in a virtual field is proposed. The aim was to measure and compare the effectiveness of the developed algorithms. Two sets of images with and without perspective effects were simulated. For images with no perspective, based on Gabor filtering and on the Hough transform, the performance of two crop/inter-row weed discrimination algorithms were tested and compared. The method based on the Hough transform is, in any case, better than the one based on Gabor filtering. For images with perspective effects only, an algorithm based on the Hough transform was tested and an extension to real images is discussed. These tests were done by a comparison between the weed infestation rate detected by these algorithms and the true one. This evaluation was completed with a crop/weed pixel classification and it demonstrated that the algorithm based on a Hough transform gave the best results (up to 90%).

Spatial and spectral methods for weed detection and localization

This study concerns the detection and localization of weed patches in order to improve the knowledge on weed-crop competition. A remote control aircraft provided with a camera allowed to obtain low cost and repetitive information. Different processings were involved to detect weed patches using spatial then spectral methods. First, a shift of colorimetric base allowed to separate the soil and plant pixels. Then, a specific algorithm including Gabor filter was applied to detect crop rows on the vegetation image. Weed patches were then deduced from the comparison of vegetation and crop images. Finally, the development of a multispectral acquisition device is introduced. First results for the discrimination of weeds and crops using the spectral properties are shown from laboratory tests. Application of neural networks were mostly studied.

Temperature imaging and image processing in the steel industry

Our aim is twofold: to present our temperature measurement system based on CCD technology, which gives a linear response versus temperature, and to display two industrial applications in which our system has been involved to optimize and characterize the process. We present a short summary dealing with temperature evaluations from radiation measurements. We consider especially the problems of the surroundings, the atmosphere, and the emissivity assumption. After selecting a value for the emissivity, we show that the use of the CCD technology enables us to obtain high spatial and temporal resolution temperature imaging, and provides further information, mainly a linear response versus temperature, which will enable real-time implementation. Our measurement system based on CCD technology is particulary dedicated to process control in the steel industry. CCD technology does not require a substantial cost and provides a lot of paramount information, enabling us to optimize some industrial applications based on accurate relative temperature measurements. We show, through two industrial examples, the results provided by our system and how it enables us to optimize the process involving knowledge and the control of temperature.

Karhunen-Loeve transform applied to region- based segmentation of color aerial images

The use of the Karhunen-Loeve transform (KLT) for region- based segmentation of aerial images by color and textural attributes is presented. Our aerial images are shown to be homogeneous color im- ages within the Karhunen-Loeve color representation space, which means they can be represented more easily and the region-based seg- mentation algorithms can be optimized. For texture analysis, the KLT is the basis of the local linear transform (LLT) and allows structural infor- mation about textures to be represented in an optimal and condensed manner. The LLT provides a system of textural analysis in the form of an adapted filter bank. We end the paper by presenting a method for merg- ing texture and color segmentations.