P. B. García-Allende

Data Processing Method Applying Principal Component Analysis and Spectral Angle Mapper for Imaging Spectroscopic Sensors

A data processing method to classify hyperspectral images from an imaging spectroscopic sensor is evaluated. Each image contains the whole diffuse reflectance spectra of the analyzed material for all the spatial positions along a specific line of vision. The implemented linear algorithm comes to solve real time constrains typical of industrial systems. This processing method is composed of two blocks: data compression is performed by means of principal component analysis (PCA) and the spectral interpretation algorithm for classification is the spectral angle mapper (SAM). This strategy, applying PCA and SAM, has been successfully tested for online raw material sorting in the tobacco industry, where the desired raw material (tobacco leaves) should be discriminated from other unwanted spurious materials, such as plastic, cardboard, leather, feathers, candy paper, etc. Hyperspectral images are recorded by a sensor consisting of a monochromatic camera and a passive prism-grating-prism device. Performance results are compared with a spectral interpretation algorithm based on artificial neural networks (ANN).

Arc-Welding Spectroscopic Monitoring based on Feature Selection and Neural Networks

A new spectral processing technique designed for application in the on-line detection and classification of arc-welding defects is presented in this paper. A non-invasive fiber sensor embedded within a TIG torch collects the plasma radiation originated during the welding process. The spectral information is then processed in two consecutive stages. A compression algorithm is first applied to the data, allowing real-time analysis. The selected spectral bands are then used to feed a classification algorithm, which will be demonstrated to provide an efficient weld defect detection and classification. The results obtained with the proposed technique are compared to a similar processing scheme presented in previous works, giving rise to an improvement in the performance of the monitoring system.

Raw Material Classification by Means of Hyperspectral Imaging and Hierarchical Temporal Memories

The recently proposed hierarchical temporal memory (HTM) paradigm of soft computing is applied to the detection and classification of foreign materials in a conveyor belt carrying tobacco leaves in a cigarette manufacturing industry. The HTM has been exposed to hyperspectral imaging data from 10 types of unwanted materials intermingled with tobacco leaves. The impact of the HTM architecture and the configuration of internal parameters on its classification performance have been explored. Classification results match or surpass those attained with other methods, such as Artificial Neural Networks (ANNs), with the advantage that HTM are able to handle raw spectral data and no preprocessing, spectral compression, or reflectance correction is required. It is also demonstrated that an optimized configuration of the HTM architecture and internal values can be derived from the statistical properties of the hyperspectral data, allowing the extension of the approach to other classification problems.

Arc-welding process control based on back face thermography: application to the manufacturing of nuclear steam generators

The possibility of reducing defects in the arc welding process has attracted research interest, particularly, in the aerospace and nuclear sectors where the resulting weld quality is a major concern and must be assured by costly, time-consuming, non-destructive testing (NDT) procedures. One possible approach is the analysis of a measurand correlated with the formation of defects, from which a control action is derived. Among others, the thermographic analysis of the weld pool and the heat-affected zone have proven to be a useful technique, since the temperature profile of the material being welded has a clear correlation with the process parameters. In this paper, we propose a control system for the submerged-arc welding (SAW) process, based on thermographic imaging of the back face of the joint being welded. In-lab experiments, with simultaneous infrared and a visible imaging, have been performed. Two image analysis techniques are proposed: tracking of the maximum temperature point of the infrared images, and morphological analysis of the visible images. In-lab welding experiments have demonstrated the feasibility of both techniques. They are able to obtain an estimation of the surface temperature and to detect the occurrence of the perforation defect, what has major application for defect detection and reduction in the joining of shell sections of nuclear steam generators.

Data processing method applying Principal Component Analysis and Spectral Angle Mapper for imaging spectroscopic sensors

A data processing method for hyperspectral images is presented. Each image contains the whole diffuse reflectance spectra of the analyzed material for all the spatial positions along a specific line of vision. This data processing method is composed of two blocks: data compression and classification unit. Data compression is performed by means of Principal Component Analysis (PCA) and the spectral interpretation algorithm for classification is the Spectral Angle Mapper (SAM). This strategy of classification applying PCA and SAM has been successfully tested on the raw material on-line characterization in the tobacco industry. In this application case the desired raw material (tobacco leaves) should be discriminated from other unwanted spurious materials, such as plastic, cardboard, leather, candy paper, etc. Hyperspectral images are recorded by a spectroscopic sensor consisting of a monochromatic camera and a passive Prism- Grating-Prism device. Performance results are compared with a spectral interpretation algorithm based on Artificial Neural Networks (ANN).

Hyperspectral Imaging for Raw Material Sorting and Processed Product Quality Control

Agricultural and industrial sectors are more competitive and quality conscious than ever before. To be profitable, industry requires equipment that would ensure pure, high-quality production, and efficient work and cost. This fact is even more important in developed countries to aid companies to defend their position and competitiveness against others where labour cost does not account for such a significant part of the overall manufacturing cost. Raw material sorters, conveyors and processing systems easy to install, energy efficient, reliable, low-maintenance and simple to adjust and control are sought. Even the inclusion of network connectivity capabilities for remote monitoring and communication is also desirable. Optical Spectroscopy (OS) becomes highly appropriate for this kind of applications because it covers all types of qualitative and quantitative analytical methods based on the interaction of light with living and non-living matter (Schmidt, 2005). For more than 200 years it has been utilized in various fields of science, industry and medicine, particularly in (bio-) chemistry, biology, physics and astronomy. OS is highly specific since each substance is discernible from all others by its spectral properties. In addition, the requirements of the samples are not particularly restrictive. Measurements of different optical parameters as a function of wavelength/energy (“spectrum”) provide valuable insights that are not, or not readily, attainable by other analytical methods. Traditional OS techniques generate information on the bulk properties of a sample or a portion taken from it (Millar, 2008). However, there are many aspects of sample properties which result from heterogeneity in composition and the monitoring of this aspect of quality is considerably more challenging. In the last few years Hyperspectral Imaging Spectroscopy (HIS) that integrates conventional imaging and moderate resolution spectroscopy, which was primarily developed for applications in remote sensing and astronomy, has been developed for laboratory use and it has even slowly transitioned into other areas as life sciences or industrial production. The key difference is that, in this technique, entire spectra are collected for each pixel within the image so its advantages in agricultural and industrial procedures such as material discrimination are obvious, i.e. reduced measurement time due to the simultaneous acquisition without the need for scanning mechanics. By assessing specific spectral features at each pixel corresponding to a material point, or by using calibration to quantify

Use of the plasma RMS signal for on-line welding quality monitoring

In this paper a new spectroscopic monitoring parameter is proposed for the on-line monitoring of welding processes, the plasma RMS signal, which is determined by considering the contribution from the spectral samples over a particular spectral window. This parameter is directly related to the heat input that can be estimated by measuring both welding voltage and current, but it exhibits a higher sensitivity to the appearance of weld defects. A comparison between the results obtained from the different spectroscopic parameters will be presented, with data from both experimental and field arc-welding tests.

Arc welding quality monitoring by means of near infrared imaging spectroscopy

The search for an efficient on-line monitoring system focused on the real-time analysis of the welding quality is an active area of research, mainly due to the widespread use of both arc and laser welding processes in relevant industrial scenarios such as aeronautics or nuclear. In this work, an improvement in the performance of a previously designed monitor system is presented. This improvement is accomplished by the employment of a dual spatial-spectral technique, namely imaging spectroscopy. This technique allows the simultaneous determination of the optical spectrum components and the spatial location of an object in a surface. In this way, the spatially characterization of the plasma emitted during a tungsten inert gas (TIG) welding is performed. The main advantage of this technique is that the spectra of all the points in the line of vision are measured at the same time. Not only are all the spectra captured simultaneously, but they are also processed as a batch, allowing the investigation of the welding quality. Moreover, imaging spectroscopy provides the desired real-time operation. To simultaneously acquire the information of both domains, spectral and spatial, a passive Prism-Grating-Prism (PGP) device can be used. In this paper the plasma spectra is captured during the welding test by means of a near infrared imaging spectroscopic system which consists of input optics, an imaging spectrograph and a monochrome camera. Technique features regarding on-line welding quality monitoring are discussed by means of several experimental welding tests.

Optimized marks for qualitative material discrimination

A method for the automatic qualitative discrimination of liquid samples based on their absorption spectrum in the ultraviolet, visible and near-infrared regions is presented. An alternative implementation of conventional spectrum matching methodologies is proposed working towards the improvement of the response time of the discrimination system. The method takes advantage of not making assumptions on the probability density function of the data and it is also capable of automatic outlier removal. Optimized marks are obtained after PCA application on the absorption spectral data. The feasibility of the algorithm has been validated with the automatic discrimination of different oil samples (seeds and olive) that have been measured with an optical fiber absorption spectroscopy set-up. The system here proposed could be easily and efficiently implemented in hardware platforms improving the system performance.

Welding diagnostics based on feature selection and optimization algorithms

In a previous paper a new approach was explored where the output parameters of a welding monitoring system based on plasma spectroscopy were the participation profiles of plasma ions and neutral atoms. They were obtained by the generation of synthetic spectra and the use of an optimization algorithm, showing correlation to the appearance of defects on the seams. In this work a feature selection algorithm is included in the model to determine the most discriminant wavelengths in terms of defect detection, thus allowing to reduce the spectral range where the synthetic spectra are generated. This should also give rise to an improvement in the overall computational performance of the algorithm. Alternatives to the use of controlled randomn search algorithms will be also explored, and the resulting model will be checked by means of experimental and field tests of arc-welding processes.