MRI study of bread baking: experimental device and MRI signal analysis

An oven dedicated to a magnetic resonance imager (MRI) was designed and constructed for continuous monitoring of the entire baking process. The general aim was to test whether response variables conventionally measured on bread loaves, such as temperature, density, and water loss, were consistent with those reported for similar products baked in classical convection ovens. MRI images acquired during baking are presented and discussed, emphasising the need to develop quantitative MRI methods allowing conversion of the MRI signal into one variable of interest, such as local density or local water content.     

Visual Servoing Based on Structure From Controlled Motion or on Robust Statistics

This paper focuses on the way to achieve accurate visual servoing tasks when the shape of the object being observed as well as the desired image are unknown. More precisely, we want to control the camera orientation with respect to the tangent plane at a certain object point corresponding to the center of a region of interest. We also want to observe this point at the principal point to fulfil a fixation task. A 3-D reconstruction phase must, therefore, be performed during the camera motion. Our approach is then close to the structure-from-motion problem. The reconstruction phase is based on the measurement of the 2-D motion in a region of interest and on the measurement of the camera velocity. Since the 2-D motion depends on the shape of the objects being observed, we introduce a unified motion model to cope both with planar and nonplanar objects. However, since this model is only an approximation, we propose two approaches to enlarge its domain of validity. The first is based on active vision, coupled with a 3-D reconstruction based on a continuous approach, and the second is based on statistical techniques of robust estimation, coupled with a 3-D reconstruction based on a discrete approach. Theoretical and experimental results compare both approaches.     

A study on local photometric models and their application to robust tracking

Since modeling reflections in image processing is a difficult task, most computer vision algorithms assume that objects are Lambertian and that no lighting change occurs. Some photometric models can partly answer this issue by assuming that the lighting changes are the same at each point of a small window of interest. Through a study based on specular reflection models, we explicit the assumptions on which these models are implicitly based and the situations in which they could fail.This paper proposes two photometric models, which compensate for specular highlights and lighting variations. They assume that photometric changes vary smoothly on the window of interest. Contrary to classical models, the characteristics of the object surface and the lighting changes can vary in the area being observed. First, we study the validity of these models with respect to the acquisition setup: relative locations between the light source, the sensor and the object as well as the roughness of the surface. Then, these models are used to improve feature points tracking by simultaneously estimating the photometric and geometric changes. The proposed methods are compared to well-known tracking methods robust to affine photometric changes. Experimental results on specular objects demonstrate the robustness of our approaches to specular highlights and lighting changes.     

Optimizing plane-to-plane positioning tasks by image-based visual servoing and structured light

This paper considers the problem of positioning an eye-in-hand system so that it becomes parallel to a planar object. Our approach to this problem is based on linking to the camera a structured light emitter designed to produce a suitable set of visual features. The aim of using structured light is not only for simplifying the image processing and allowing low-textured objects to be considered, but also for producing a control scheme with nice properties like decoupling, convergence, and adequate camera trajectory. This paper focuses on an image-based approach that achieves decoupling in all the workspace, and for which the global convergence is ensured in perfect conditions. The behavior of the image-based approach is shown to be partially equivalent to a 3-D visual servoing scheme, but with a better robustness with respect to image noise. Concerning the robustness of the approach against calibration errors, it is demonstrated both analytically and experimentally.     

Determination of the lean meat percentage of pig carcasses using magnetic resonance imaging

The purpose of Workpackage 3 of the European Eupigclass project was to test indirect methods of measuring the lean meat percentage of a carcass that would be less costly, at least as accurate and more consistent than dissection. Magnetic resonance imaging was one of the three indirect methods tested to measure the lean meat weight and the lean meat percentage of pig carcasses, the other methods being X-ray CT and vision techniques. One hundred and twenty carcasses from three different genotypes and from both sexes were slaughtered. The left parts of the carcasses were fully dissected and the right parts were investigated with an indirect method using a 1.5T MRI system. The acquisition protocol was chosen to give an optimized contrast between fat and muscle tissues. Two different approaches, image segmentation and PLS regression, were used to extract information from the images. Automatic image segmentation was performed to quantify the volume of muscle in the images and gave a standard error of prediction using a linear regression with the dissection of the left half carcasses of 586 g and 1.10% for lean meat weight and lean meat percentage, respectively. PLS regression using the signal intensities histograms gave an estimation error of 465 g for lean meat weight. These results showed that MRI could be used in place of full dissection for authorizing and monitoring classification equipment of pig carcasses.     

Quantification of muscle, subcutaneous fat and intermuscular fat in pig carcasses and cuts by magnetic resonance imaging

The aim of this study was to determine the suitability of magnetic resonance imaging (MRI) to predict tissue composition of pig carcasses and cuts. Twenty-four pig carcasses were cut into the four primary cuts that were analyzed with a low field MRI imager before a total dissection. Images were then processed to identify and quantify pixels representing muscle, subcutaneous fat and intermuscular fat fractions. MRI provided a good prediction of muscle content in cuts and carcasses, with R² ranging from 0.970 to 0.997. The prediction was slightly less accurate for total fat (0.951 ? R² ? 0.986) or subcutaneous fat (0.918 ? R² ? 0.994). Finally, the prediction of intermuscular fat content in considering intermuscular fat classified pixels was acceptable only for the belly (R² = 0.837).     

Estimation of poultry breastmeat yield: magnetic resonance imaging as a tool to improve the positioning of ultrasonic scanners

Magnetic resonance imaging (MRI) was used to reconstruct three-dimensional breast muscle volume of 30 broilers and to locate the most suitable cross-sections to estimate breast muscle yield by ultrasonic scanner. The high accuracy of the determination of the breast muscle yield (R(2)=0.92) from the volume calculated by the sum of 6 mm-spaced MRI transverse images justified the choice of MRI as a reference method. Treatment of the images showed that it was possible to obtain acceptable breast meat yield prediction by MRI from a combination of two or three muscle transverse cross-section area measurements. It also showed that the need to find interfaces reflecting ultrasound is a considerable handicap for optimizing the ultrasonic technique. The oblique echotomographic plane crossing the coracoid bone lengthwise and the fore part of the breast bone appears to be the most appropriate to improve significantly the determination provided by the transverse image situated at the fore part of the breast bone.