Pierre Bonton

Development of a semi-automatic system for pollen recognition

A semi-automatic system for pollen recognitionis studied for the european project ASTHMA. The goal of such a system is to provideaccurate pollen concentration measurements. This information can be used as well by thepalynologists, the clinicians or a forecastsystem to predict pollen dispersion. At first,our emphasis has been put on Cupressaceae, Olea, Poaceae and Urticaceae pollen types. The system is composed of two modules: pollengrain extraction and pollen grain recognition. In the first module, the pollen grains areobserved in light microscopy and are extractedautomatically from a pollen slide coloured withfuchsin and digitized in 3D. In the secondmodule, the pollen grain is analyzed forrecognition. To accomplish the recognition, itis necessary to work on 3D images and to usedetailed palynological knowledge. Thisknowledge describes the pollen types accordingto their main visible characteristerics and tothose which are important for recognition. Somepollen structures are identified like the porewith annulus in Poaceae, the reticulum in Oleaand similar pollen types or the cytoplasm inCupressaceae. The preliminary results show therecognition of some pollen types, likeUrticaceae or Poaceae or some groups of pollentypes, like reticulate group.

A guidance-assistance system for agricultural vehicles

Abstract This article presents a guidance-assistance system for agricultural machines. It is based on analysis of the vehicle’s environment by image processing to deduce a control law in the image space. Two algorithms of image processing and two control laws are presented. It presents some results of crop edge detection and control of the vehicle’s trajectory in several conditions. These results show the feasibility of such a project and the necessity to know the reliability of the image processing results to secure the viability of the complete system.

An original correlation and data fusion based approach to detect a reap limit into a gray level image

Real time vision systems in an outdoor environment involve an increase of the algorithm complexity in order to solve the application with an acceptable autonomy. In our application of help guidance for agricultural mobile machines, the vehicle has to follow the limit between a mowed and an unmowed natural zone. After the extraction of the image by a CCD camera, a low level algorithm computes some luminance and texture parameters for each elementary site. We define a correlation function for each parameter. The original feature of our work is a geometrical characterization of the correlation functions, as well as performs belief functions associated with each parameter. The uncertainty notion, expressed by the use of the theory of evidence allows the control of the reliability of the estimation. This approach is very important in outdoor environment where the system can be confronted in many situations.

Learning Method for In-Situ Soil Classification Based on Texture Characteristics

Knowing the nature of a soil through its classification is a key step in understanding its behaviour. In addition to its mechanical properties, information on soil type is important for forecasting material behaviour. This requires laboratory tests that are often long and expensive. Our approach is a development of in-situ identification tests used to complement “blind” mechanical tests. A new method is proposed for obtaining in-situ soil classification based on the use of geo-endoscopy and image analysis. This technique is based on the computation of texture features in geoendoscopic images and, more particularly, second order statistical features. After a presentation of the goals of this research, this article presents the methodology proposed and the learning steps performed to achieve these goals. The second part describes the testing and selection of major textural features according to their performances and their discriminating capacities. In the last part the classification results obtained with this methodology are presented. Soil image classification has been tested successfully and achieves accuracy of over 80 % in classifying natural soil samples. Because of the fixed magnification of the camera, results presented here are strongly linked to this magnification.

Unsupervised Regions Segmentation: Real Time Control of an Upkeep Machine of Natural Spaces

An original image segmentation based on a Markovian modeling of a set of four parameters is presented. The application performed demonstrates the strength of an algorithm using texture analysis of natural scenes. A splitting limit is obtained which is going to become the basic primitive in order to hook up a mower robot. To obtain a real time application we also present a simple parallelization of the algorithm and a control servoing.