Potential of Hyperspectral Imaging for Quality Control of Polymer Blend Films

Abstract

A visible−near-infrared (VIS-NIR) hyperspectral imaging sensor is proposed for online nondestructive monitoring of polymer film blends produced by extrusion blowing. Such a sensor provides a much higher spatial resolution of analysis compared to probes or probe arrays, which should help in detecting smaller localized defects inducing poor film quality (mechanical, optical, barrier properties) at a macroscopic scale. Multiresolutional multivariate image analysis (MR-MIA) was used to extract those spectral and textural features from images that are the most highly correlated to film composition distribution and mechanical properties. When applied to a low-density polyethylene/polystyrene (LDPE/PS) blend produced under various operating conditions, the imaging sensor was able to monitor the two dominant effects on mechanical properties: (1) the effect of film composition (raw materials) and (2) the stretching effect caused by postextrusion conditions (process).

Visible and near-infrared light transmission: A hybrid imaging method for non-destructive meat quality evaluation

“…To do this, a hyperspectral imaging system was used to obtain the spectra of marbling and muscle of a sample of beef meat. This system [22][23], consists of two linear (line-scan) spectroscopic imaging systems, effectively covering the visible and NIR spectra (400 to 1700 nm). We used this system to measure the absorbance spectra of the marbling and muscle of a sample of beef meat.…”

“…NIR light analysis provides valuable quantitative information for applications in several fields. For meat, several research projects have demonstrated that it can be used to measure moisture and marbling [21,22,23].…”

Wavelet texture analysis in process industries

“…This is probably the most frequently cited method for texture analysis and many variations of GLCM such as texture unit [20] and neighboring grey level dependence matrix (NGLDM) [21] have also been proposed. The GLCM method in general gives relatively good results in recent applications [13].…”

“…On the contrary to typical manufacturing processes where images always provide a scene of objects with pre-defined size, shape, alignment, and so on, many processes in (chemical) process industries provide images where stochastic nature of the visual scene is dominant. In such processes, some hard-to-define outer appearance of products or processes is mostly major concern: color and morphology of froth in flotation processes [8,9], aesthetics of engineering-stone countertops [10], random visible pattern in injection-molded plastic panels [11], coating uniformity of medication tablets in an industrial coater [12], spatio-temporal variations of properties of polymer composite [13], quality control of paper formatoion [14] are just a few examples. In the example of surface roughness of rolled steel sheets [15], the quality of a steel sheet is related to the number and severity of pits on its surface.…”

Characterization of PLA-talc films using NIR chemical imaging and Multivariate Image Analysis techniques

“…However, most of the work in this area focused on the qualitative analysis of plastic films [19,22] or are limited to quantify the concentration of specific constituents or contaminants in the films [18,23] which is not enough to meet the quality control objectives. A very limited number of investigations were published on polymer film monitoring and image-based properties prediction, such as crystallinity, mechanical and gas barrier [63], [70], [98].…”