The infra-red, absorption of milk and its main constituents has been recorded and it has been found possible to determine quantitatively the fat, protein and lactose contents from measurements of the intensities of the respective absorption peaks at 5-73, 6-46 and 9-6/tm. Solids-not-fat (SNF) content can be obtained from a single intensity measurement at 7-9 /im. where the fat shows no attenuation and where the protein and lactose extinction coefficients are approximately equal. Photometric errors due to natural variations in fat globule size distribution can be eliminated by prior homogenization. Based on these observations, an automatic infra-red milk analyser (IRMA) has been constructed, which can determine the fat, protein, lactose and SNF in a pre-warmed milk sample in under 1 min. Preliminary tests carried out on 50 milk samples from individual cows, representing 3 different breeds, gave standard deviations from the chemical analyses for the percentages of fat, protein, lactose and SNF of 0-10, 0-07, 0-07 and 0-19, respectively. The standard deviations for the percentages of fat and SNF on 60 farm bulk milk samples were found to be 0-08 and 0-12. This paper presents the basic principles and preliminary results of a rapid method for the quantitative analysis of milk by means of infra-red absorption. The method has already been reported briefly (Goulden, 1961a) and is covered by a patent vested in the National Research Development Corporation (Goulden, 19616). A commercial instrument employing this principle is now being developed and its constructional features have been described by Goulden, Shields & Haswell (1964). THEORETICAL Infra-red absorption spectroscopy is commonly used for the quantitative analysis of multicomponent systems, but the high water content and radiation scattering properties of milk make it a particularly difficult system for analysis by this method. As can be seen from the spectra shown in Fig. 1, even thin films of water absorb strongly throughout the 2-10 /iva wavelength region. The water absorption bands are so intense that it is almost impossible to discern bands due to other milk constituents. When a double-beam spectrometer is used to cancel out automatically the water absorption, the characteristic bands of the other components become clearly visible as in the spectra shown in Fig. 2. These bands arise from particular components of milk and when recorded as optical densities (absorbancies), the band intensities are proportional to the concentration of each of the major constituents.

Routine EBSD analysis typically applies a 2D Hough transform technique. This requires collection and background correction of the Kikuchi pattern. A Hough transform is applied to highlight and locate the Kikuchi bands. These bands are revealed as local peaks in the resulting Hough space and their corresponding maxima are found automatically, locating the position and direction of the Kikuchi bands and the corresponding diffraction planes. In this primary band detection routine each maximum is represented by a (rho, theta) position in Hough space. In the subsequent indexing the interplanar angles are matched against a database of known phases, and the phase and orientation of the crystal is determined.Precision and accuracy of this primary band detection is essentially limited by the precision of identifying and locating the Kikuchi bands, which is in turn limited by the Hough transform. Therefore, the precision in the orientation measurement can be improved by increasing the resolution of the Hough transform and the resulting Hough space; however this results in a significant increase in calculation time. In addition, the Hough transform assumes that the Kikuchi bands are straight lines, whereas in reality the Kikuchi bands in a pattern have varying widths and curved (hyperbolic) edges. Therefore the conventional 2D Hough transform method introduces a systematic error which cannot be overcome through increased Hough space resolution.An alternative method [1] describes modifying the Hough transform so that it identifies hyperbolic curves rather than straight lines. This modified Hough transform applies a 3D space where the third dimension is given by the hyperbola variable. This method eliminates the systematic error of the conventional Hough, but it involves a large three dimensional calculation which makes it computationally slow. This paper will present a new analysis routine for EBSPs, which improves the accuracy of band detection and is achievable in real time applications. The method uses the primary band detection as its foundation, and then applies an iterative secondary band detection to improve upon the accuracy of the primary band detection.This accurate determination of band positions enables the calculation of a more accurate orientation matrix and thus improves general EBSD analysis, e.g. ability to distinguish between two phases with a very close crystal structure; or in characterizing low angle boundaries. In addition, this refinement will deliver an accurate solution when the primary band detection is impaired by the presence of excess and deficit lines.The mean angular deviation (MAD) is a measure of the fit between the Kikuchi bands in the measured pattern and the solution. It can be used as figure of merit for the accuracy of band detection and therefore also for phase identification. Many applications of EBSD consider misorientation which is also related to the accuracy and precision of band detection. 724

SummaryThe homogenization of milk and cream with a piston-type homogenizer has been studied. An examination has been made of the influences upon the mean fat globule diameters of changes in milk fat content, flow rate and homogenization temperature. The results are discussed in terms of the variations in the parameters q and P0 of the empirical equation d = (P0/P)q relating the globule diameter d to applied pressure P.

No abstract

No abstract