Maik Rosenberger

A novel 3D multispectral vision system based on filter wheel cameras

The combination of the multispectral imaging and the 3D imaging in one system would render possible varieties of applications in different fields. In this work a method is proposed for the registration of multispectral data and 3D point cloud, in which the number of spectral bands is increased by combining different multispectral images through 3D information. Based on this method, a 3D multispectral imaging system consisting of two filter wheel cameras and a digital projector is developed. This system can provide maximally 23 spectral bands with a short acquisition duration of ca. 1.5 seconds and achieve a data fusion accuracy of 15 μm for a measurement field of 128 mm × 145 mm. Furthermore, there exists a significant potential to extend the spectral range and raise the spectral resolution by integrating additional cameras with the proposed data fusion principle.

Development and characterization of a high speed linear- moving-stage for multispectral measurements

Multispectral imaging is one of the key topics in the upcoming years, a lot of measurement tasks can be solved using the additional spectral information. For the separation of these additional information bands plenty technologies were developed in the last decades. All of them have characteristic advantages and disadvantages. One working principle is the use of a filter wheel containing different filters between the image sensor and the lens. The additional element inside the light propagation path as well as the lens imperfections lead to chromatic aberrations. To compensate the resulting focus error some investigation were made. One option is the movement of the image sensor stage to a better focus position. Therefore high precision moving stage with high movement speed are characteristic parameters which have to be considered during the construction. In this paper a linear motor stage paired with two high precision linear bearings were connected together to achieve the demanded requirements. For the evaluation of this new approach, several measurements determining the dynamic as well as the static specifications were made to proof the achieved characteristics and will be shown in the paper. With this approach, a miniaturised precision system inside the multispectral imager a precision repeatability of less than four microns can be achieved.

Nearfield sensing and actuation for multispectral imaging systems

The sharpness of edge transition is one of the major requirements in optical sensing especially in the field of image processing. Therefore a lot of algorithms were assessed in the field of digital photography. In the field of industrial image processing especially multispectral imaging different focus positions are needed to get an optimal focus position for every sampling channel. Therefore a control loop was developed which uses an inductive sensor to measure the actual position of an image sensor. With this information a voice coil were activated to actuate the sensor into the optimal position. For a correct operation a resolution of less than fifty microns have to be reached. The paper presents the construction as well as the electronic and sensor actor design. First evaluations demonstrating a proof of concept and will also discussed in the paper.

Geometrical and spectral data stitching for combining hyperspectral imaging systems

Abstract One way to increase the amount of information acquired via hyperspectral imaging and therefore to increase the possibility of data analysis is combining the spatial and spectral information of hyperspectral data sets. The aforementioned data sets are obtained by cameras covering different spectral ranges. The purpose of this article is to develop an algorithm which is able to combine two data sets acquired by two hyperspectral pushbroom imagers, covering the visible (VIS) and the near infrared (NIR) wavelength range. Initially, the effect of optical aberrations, as well as errors via the image registration were examined. Subsequently a correction algorithm for both the optical aberration and the image registration is elaborated.

A novel algorithm for bad pixel detection and correction to improve quality and stability of geometric measurements

An algorithm for detection and individually substitution of bad pixels for further restoration of an image in the presence of such outliers without altering overall image texture is presented. This work presents three phases concerning image processing: bad pixel identification and mapping by means of linear regression and the coefficient of determination of the pixel output as a function of exposure time, local correction of the linear and angular coefficients of the outlier pixels based on their neighbourhood and, finally, image restoration. Simulation and experimental data were used as means of code benchmarking, showing satisfactory results.

Multispectral edge detection algorithms for industrial inspection tasks

Multispectral imaging technology enables a lot of possibilities in industrial imaging and automation. Especially parts which are complicated for inspection can be inspected using more than one or three wavelength channels. One challenge is the use of all channel information in the multispectral image for the edge detection. For this purpose different approaches were investigated. The results of the investigations in different edge detection models were applied for the segmentation of specific boundaries on composite pipes. With the presented multispectral imager and the investigated methods in multispectral image processing, the different layers on the pipe were segmented with subpixel accuracy as well as with pixel accuracy. Furthermore the challenges for illumination correction as well as geometric correction were discussed.

Automatic detection of fused tungsten carbides for the characterization of welding process

Abstract This paper represents an imaged-based method for the quantitative characterization of the dissolution behaviour of fused tungsten carbide in the welding process. For the metallographic investigations based on microscopic images, a hybrid image segmentation procedure for the detection of hard phases is proposed to quantify their amount and distribution in the matrix material. With the help of this tool, the dependence of hard phase characteristic on the welding parameters was investigated.

Adaptive test bench for characterizing image processing sensors

The Group for Quality Assurance and Industrial Image Processing has built a new experimental setup to characterize cameras and image sensors according to EMVA1288 standard. Next to the investigation of SLN (Sensitivity-Linearity- Noise) and spatial non-uniformities, the new test bench also provides an examination of the temperature dependence of sensors. The temperature dependent dark current produces an undesirable signal which affects the image quality negative and thus has to be known. It is caused by thermally induced electrons and increases linearly with exposure time as well as exponentially with temperature. To measure the dark current, it is necessary to vary and determine the temperature of the sensor. This was made possible by a climate chamber with a Peltier element, which enables a heating and cooling of the camera. An infrared sensor allows a contact-free detection of the actual camera temperature. Furthermore, the light source was improved for the new test bench. With the installed custom light source and integration sphere a homogeneous irradiation up to 97% is ensured. This way better results in tests were achieved. The light source with variable filter housing enables the use of monochromatically light in a wavelength range of 350 – 1700nm. A live monitoring of the irradiation during the image capturing is possible. A MATLAB script assists in the configuration of the camera, the measurement and the data storage. The user is guided step by step through the program. At the end of the measurements an automated evaluation follows, which illustrates graphs and parameters in a streamlined and print-ready format.

Wide range UV irradiation system for imaging reflection spectroscopy

Reflection spectroscopy is a promising measuring principle to get measurement data of surface signatures from objects and materials. A large field of applications, in inline or outline testing tasks, can be handled with it. The investigation deals with investigations regarding development of a broadband UV radiation source and the integration in a hyperspectral push broom imaging system using standard spectral radiation sources. Comprehensive market research has shown that there is no disposability of a UV radiation source, which emits a nearly continuous radiation in the spectral range of 220 up to 400 nanometers. Therefore, a second idea is to combine various gas discharging lamps to receive a nearly continuous spectrum in that area of the UV radiation. A very important and interesting circumstance of using standard spectral lamps as a radiation source for push-broom-imaging is the fact, that the active zone of the plasma discharge has a line-like contour, which can assumed as an axially expanding point light source. Therefore, it is only necessary to project this for getting the needed radiation field in the framework of reflection spectroscopy. The aim is to determine which different gases of a standard range of spectral light sources are suitable. Another approach is the investigation of the effect that modifiable parameters – such as pressure or a time modulated energy source - might have to the spectral behavior of the differing radiation sources and the reflected light from the test object as well. Physical models of high intensity discharging light sources are well proved, but the concept of light addition, to get a special required spectral distribution, may cause some specific problems that must be solved. It is also a goal to evaluate, in which conditions classic models work and how far the can be used.

Programmable system on chip implementation of a principal component analysis for preprocessing of multispectral image data acquired with filter wheel cameras

The acceleration of the acquisition of spectral images and their processing is important for the acceptance of these measurement methods in quality assurance and inspection. A frequently used preprocessing step is the Principal Component Analysis (PCA). It is used in variations, for example, for segmentation, spectral decomposition or data compression. The presented implementation calculates the PCA for the 12 spectral image channels of a filter wheel camera parallel to image acquisition. This includes the determination of the covariance matrix, the calculation of the main components and the transformation of the data. The parallel processing during the sequential imaging acquisition is performed on a System-on-a-programmable-chip (SoPC) Xilinx Zynq-7000 directly within the camera. The algorithm is partitioned into hard and software components and implemented in the field programmable gate array (FPGA) fabric as well as the ARM processor core firmware of the SoPC. In order to ensure the steps of the image acquisition chain in addition to the calculation, the system was implemented as an asymmetric multiprocessing system (AMP) with individual processors. For additional acceleration under static conditions (e.g. continuous testing in the manufacturing process), the feature vector can be stored as a calibration value. The calculation is reduced to the transformation of the data.