Sebastian Budzan

Face and eyes localization algorithm in thermal images for temperature measurement of the inner canthus of the eyes

Abstract In this paper, a novel algorithm for the detection and localization of the face and eyes in thermal images is presented, particularly the temperature measurement of the human body by measuring the eye corner (inner canthus) temperature. The algorithm uses a combination of the template-matching, knowledge-based and morphological methods, particularly the modified Randomized Hough Transform (RHT) in the localization process, also growing segmentation to increase accuracy of the localization algorithm. In many solutions, the localization of the face and/or eyes is made by manual selection of the regions of the face and eyes and then the average temperature in the region is measured. The paper also discusses experimental studies and the results, which allowed the evaluation of the effectiveness of the developed algorithm. The standardization of measurement, necessary for proper temperature measurement with the use of infrared thermal imaging, are also presented.

Fusion of Visual and Range Images for Object Extraction

This paper proposes a fused vision system using range laser scanner and visual camera for object extraction in mobile systems. Fusion of information gathered from different sources increases the effectiveness of the small objects detection in different scenario, e.g. day, night, outdoor, indoor, sunny or rainy weather. First of all, the algorithm for color images is proposed for extracting objects from the scene. The labelled objects are divided into two classes: background and obstacles, based on the morphological operations and segmentation method. Range laser measurement system is used regardless of the visual images classification to the obstacle and non-obstacle only. After that the size (width, height, depth) of the labelled objects are determined. Then the knowledge rules have been used to classify objects into separate three obstacle classes: small, medium and large.

Grain Size Determination and Classification Using Adaptive Image Segmentation with Shape-Context Information for Indirect Mill Faults Detection

In this paper, the authors described methods of material granularity evaluation and a novel method of grain size determination with inline mill device diagnostics. The mill diagnostic can be carried out with vibration measurements, machine vision or infrared imaging. Milling process is an extremely energy- and cost-consuming process, thus diagnostics process should be performed with high efficiency. Method proposed in this paper is based on the online examination of the final product during the milling process using real-time digital images. Determination of the total number of the grain, size of each grain, also classification of the grains is main goal of proposed method. In the proposed method, the visible camera with lightning mounted at two assumed angles has been used, what increases the grain detection process. Proposed method uses an adaptive segmentation to match correctly the grains, the information about particles shape and context is used to optimize the grain classification process in the next step. Finally, during image processing, the simple rule-based method is used to obtain information about overall quality of the product and possible faults in the milling process based on the evaluated parameters.

Noise Reduction in Thermal Images

In this cognitive work we focused on investigation of some filters used for image processing in application for noise removal in IR images. In IR imaging the choice of filter depends mainly on the purpose of the processing, e.g. detection of small objects in complex images, edge and contour detection or removal of non-uniformity of the detector array. The performance of the selected noise reduction filters was evaluated using PSNR and RMSE quality measure. The results are shown only for few images from our database which contain over 2000 of IR images.

RT/FPGA implementation of the IEEE 1451 standard in sensors for machine conditioning systems

IEEE 1451.4 standard introduced the full “plug and play” for sensors and actuators thanks to the Transducer Electronic Datasheet (TEDS). The only requirement for the measuring system is compatibility with 1451.4. The advantage of this standard among all series of IEEE 1451 is that the Network Capable Application Processor (NCAP) is not necessary. This caused the IEEE 1451.4 to become the most popular and widely available on the market. This paper presents the implementation of the IEEE 1451.4 standard in custom made vibration transducers, based on MEMS accelerometers. The transducers were calibrated and the results of calibration are stored in TEDS added to the sensor. Then the transducers with TEDS were utilized in an experimental machine conditioning system based on embedded RT/FPGA platform.

Rotating Machinery Diagnostics Based on Fusion of Infrared and Vibration Measurements

Systems for on-line vibration diagnostics of rotary machines requires the installation of vibration sensors on practically each bearing. For a typical machine there are at least a couple of bearings to be monitored and for many machines to be monitored it generates considerable costs arising from the price of sensors, diagnostic modules, software and wiring. In this paper we discuss the possibility of reduction in the number of vibration measurement points, with occasional inspection with IR (Infra-Red) camera, which globally could reduce the system costs without compromising the quality of the machine diagnostic. A measurement stand modeling operation of a simple machine driven by an electric motor and a typical industrial vibration diagnostics system together with a thermal imaging camera were used for this purpose. It allows controlled induction of various defects that result in vibrations. Vibrations was measured with properly parameterized system for on-line vibration diagnostic, separately for each bearing. The research involved performing a few experiments, which modeled a typical defect in rotating machines: quasi static unbalance of the shaft, outer race defect of the bearing and cage defect of the bearing. The fusion of obtained IR and vibration information was discussed.

Human Detection in Low Resolution Thermal Images Based on Combined HOG Classifier

The human detection in real environment is important task of the computer vision, especially if we take into account thermal imagery. Most of the recent methods are based on the low-level features or body parts detection or combination. Method proposed in this paper uses combination of modified Histogram of Oriented Gradients (HOG) with detection of the human head. The minimal distance classifier has been used to improve the reduction of the human candidates process. The experiments have been performed on thermal images taken in real environment in different scenario such as missing body parts, overlapped people, different pose, far and near distance to the human, small groups of people, large groups of the people. The performance of the proposed algorithm has been evaluated using Precision and Recall quality measure with comparison to the selected reference methods.