In the past three years, deep convolutional neural networks (DCNNs) have achieved promising performance in detecting skin cancer. However, improving the accuracy and efficiency of the automatic detection of melanoma is still urgent due to the visual similarity of benign and malignant dermoscopy. There is also a need for fast and computationally effective systems for mobile applications targeting caregivers and homes. This paper presents the You Only Look Once (Yolo) algorithms, which are based on DCNNs applied to the detection of melanoma. The Yolo algorithms comprise YoloV1, YoloV2, and YoloV3, whose methodology first resets the input image size and then divides the image into several cells. According to the position of the detected object in the cell, the network will try to predict the bounding box of the object and the class confidence score. Our test results indicate that the mean average precision (mAP) of Yolo can exceed 0.82 with a training set of only 200 images, proving that this method has great advantages for detecting melanoma in lightweight system applications.

Autoregressive modelling techniques such as multi-channel linear prediction are widely used for applications such as coding, dereverberation and compression of the speech signals. State of the art multi-channel linear prediction methods do not take into account the locations of the microphones and assume single distance compact microphone arrays. In this paper a spatially modified multichannel autoregressive compression and coding method is proposed and successfully tested in order to adapt the standard multi-channel method to the virtual reality and immersive video conferencing applications where the microphones can be meters away from each other. The proposed method estimates the spatial distances between each microphone and the source to optimise the joint compression of the signals recorded within a wide area. The results suggest that the proposed method outperforms the standard multi-channel compression and coding when applied to the ad-hoc scenarios.

Coating paperboard enhances printability and optical quality of the product as well as other important properties like packaging performance and shelf-life. To obtain high quality products a quality control of the manufacturing process requires identifying those manufacturing parameters that affect it. Roughness measurement and characterization of coating thickness are examples of these control parameters. Optical instruments measuring these quantities range from laboratory equipment to in-line and on-line sensors. However, the variety of instruments and sometimes misunderstanding of their limitations generate uncorrelated measurements, which are no longer valid to their comparison. The new ISO 25178 standard for surface texture provides guidelines to metrologists to address some this problem. Here, we report a case study for surface characterization of high quality printing polyethylene (PE) coated paperboard for high quality printing, where surface roughness is a key parameter. Two imaging methods to create topographic measurements will be compared, i.e. a confocal chromatic microscope and a scanning electron microscope (SEM). For the latter, stereo photogrammetry is used and 3D topographic profiles are obtained from Alicona MeX software. Leach and Haitjema [Leach, R., & Haitjema, H. (2010). Bandwidth characteristics and comparisons of surface texture measuring instruments. Measurement Science and Technology, 21(3), 032001] addressed the topic on how to design comparisons when using different instruments for areal texture measurement. We use their bandwidth matching concept, since it provides an extension to the ISO 25178 guidelines, ensuring that the instrumentation used to characterize the samples are within its measuring limits and further analysis of the results can be correlated. It is important to adopt a good metrology practice in order to translate these parameters into our future work. We expect to extend these findings into a real-time optical sensor, which later can be implemented in an industrial manufacturing environment for high optical quality paper and paperboard.

X-ray imaging has been used extensively in the manufacturing industry. In the paper and paperboard industry X-ray imaging has been used for measuring parameters such as coat weight, using mean values of X-ray absorption inline in the manufacturing machines. Recently, an interest has surfaced to image paperboard coating with pixel resolved images showing material distribution in the coating on the paperboard, and to do this inline in the paper machine. Naturally, imaging with pixel resolution in an application where the paperboard web travels with velocities in the order on 10 m/s sets harsh demands on the X-ray source and the detector system to be used. This paper presents a scanning imaging method for single photon imaging systems that lower the demands on the source flux by hundreds of times, enabling a system to be developed for high velocity industrial measurement applications. The paper presents the imaging method, a discussion of system limitations, simulations and real measurements in a laboratory environment with a moving test object of low velocity, all to verify the potential and limits of the proposed method.

Recent advancements in the grating interferometer based Phase Contrast X-ray Imag- ing (PCXI) technique enables high quality dark-field images to be obtained using conventional X-ray tubes. The dark-field images map the scattering inhomogeneities inside objects. Since, the dark-field image is constructed by considering only those photons which are scattered while pass- ing through the objects, it can reveal useful information about the object inner structures, such as, the fibre structures inside paperboards.

The end-use performance of paperboards, such as the printing quality and the stiffness de-pends on the uniformity in the thickness and the structures of the coating layer of the paperboards. The uniformity in the coating layer is determined by the coating techniques, the coating materials and the topography of the base sheet. In this article, the dark-field signals from four paperboard samples with different quality indices are analysed. The isotropic and the anisotropic scattering coefficients for all of the samples have been calculated. Based on the correlation between the isotropic coefficients and the quality indices of the paperboards, a new method for paperboard quality measurement has been suggested.

A new prototype device has been developed based on a laser triangulation principle to measure online surface topography in the paper and paperboard industries. It characterizes the surface in a wide spatial scale of topography from 0.09-10 mm. The prototype's technique projects a narrow line-of-light perpendicularly onto the moving paper-Web surface and scattered reflected light is collected at a low angle, low specular, and reduced coherent length onto the CCD sensors synchronized with the laser sources. The scattering phenomenon determines surface deviations in the z-direction. The full-width, at half-maximum of a laser line in cross section is sensitive in computation of the surface topography. The signal processing aspect of the image processing, for example, threshold and filtering algorithms are also sensitive in estimating the accurate surface features. Moreover, improper light illumination, intensity, reflection, occlusion, surface motion, and noise in the imaging sensor, and so forth, all contribute to deteriorate the measurements. Optical techniques measure the surface indirectly and, in general, an evaluation of the performance and the limitations of the technique are both essential and challenging. The paper describes the accuracy, uncertainty, and limitations of the developed technique in the raw profiles and in terms of the rms roughness. The achieved image subpixel resolution is 0.01 times a pixel. Statistically estimated uncertainty (2σ) in the laboratory environment was found 0.05 μm for a smooth sample, which provides a 95% confidence level in the rms roughness results. The depth of field of the prototype is ~2.4 mm.

Real time surface topography measurement in the paper and paperboard industries is a challenging research field. The existing online techniques measure only a small area of paper surface and estimate topographical irregularities in a narrow scale as a single predictor. Considering the limitations and complications in measuring the surface at high speed, a laser line triangulation technique is explored to measure surface topography in a wide scale. The developed technique is new for the paper and paperboard application that scans a line onto the paper-web surface up to 210 mm in length in the cross machine direction. The combination of a narrow laser linewidth imaging, a subpixel resolution, and the selection of a unique measurement location has made it possible to measure roughness and simultaneously characterize paper surface topography from 0.1 to 30 mm spatial wavelength. This spatial range covers wide scale surface properties such as roughness, cockling, and waviness. The technique clearly distinguishes and characterizes the surface of newspaper, and lightweight coated, coated, and uncoated paperboard in real time during the paper manufacturing process. The system temporal noise for the average roughness is estimated as 37 dB. The signal to noise ratio found is from 5.4 to 8.1 in the short spatial wavelength up to 1 mm, whereas it is more than 75 in the long spatial wavelength from 5 to 10 mm.

Quality control is an important issue in the paperboard industry. A typical sheet of paperboard contains a core of cellulose fibers [C6H10O5], coated on one or both sides with layers of calcium carbonate [CaCO3] or Kaolin [Al2Si2O5(OH)4]. One of the major properties of a good quality paperboard is the consistency of the expected ratio between the thickness of the core and the coating layers. A measurement system to obtain this ratio could assist the paperboard industry to monitor the quality of their products in an automatic manner. In this work, the thicknesses of the core and the coating layers on a paperboard with coating layer on only one side were measured using an X-ray imaging technique. However, the limited spectral and spatial resolution offered by the measurement system being used led to the measured thicknesses of the layers being lower than their actual thicknesses in the paperboard sample. Suggestions have been made in relation to overcoming these limitations and to enhance the performance of the method. A Monte Carlo N-particle code simulation has been used in order to verify the suggested method.

In the paper industry, surface topography is the essence of both paper and paperboard, and accurate topographical measurements are equally essential in order to achieve a uniform smooth surface. The traditional laboratory methods measure only a few samples from the entire tambour and there are other obvious limitations to this approach. Online measurements may be of significant value to improve the surface quality throughout the production. Roughness is one of the topography components and the majority of techniques measure paper by means of a single predictor of average roughness, R _a which is inadequate in providing a comprehensive characterization of the surface. Measurements, in a wide range ofwavelengths, can characterize topography components such as roughness, waviness, cockling, etc. Online measurements were taken for various grades of 8 paper reels, containing the wireside and topsides for newspaper, and uncoated and coated sides of paperboards. Their surfacecharacterization, in the spatial wavelength spectrum, from 0.1 to 10 mm was obtained. This article presents the online characterizationresults which have efficiently distinguished the surfaces of same family materials including the edge and the middle position reels of fine coatedpaperboard. Online measurements were taken, at Iggesund Paperboard Pilot Coater in Sweden, by using a recently developed OnlineTopography (OnTop) device which is based on the principle of light triangulation. © 2012 Elsevier B.V. All rights reserved.