Arto Kaarna

Lappeenranta – Lahti University of Technology LUT · Department of Mathematics and Physics

Deep learning models have the potential to advance the short-term decision-making of electricity market participants and system operators by capturing the complex dependences and uncertainties of power system operation. Currently, however, the adoption of global deep learning models for multivariate energy forecasting in power systems is far behind...

Forecasting the state-of-charge changes of battery energy storage, anticipated from a provision of different services, can facilitate planning of its market participation strategy and leverage the maximum potential of its energy capacity. This paper provides a performance comparison study of multiple decision-tree and data-driven machine learning m...

Unsupervised classification is a crucial step in remote sensing hyperspectral image analysis where producing labeled data is a laborious task. Spectral Clustering is an appealing graph-partitioning technique with outstanding performance on data with non-linear dependencies. However, Spectral Clustering is restricted to small-scale data and neither...

Unsupervised classification is a crucial step in remote sensing hyperspectral image analysis where producing training labelled data is a laborious task. Hyperspectral imagery is basically of high-dimensions and indeed dimensionality reduction is considered a vital step in its preprocessing chain. A majority of conventional dimensionality reduction...

This paper proposes an outlier robust geodesic K-mean algorithm for high dimensional data. The proposed algorithm features three novel contributions. First, it employs a shared nearest neighbour (SNN) based distance metric to construct the nearest neighbour data model. Second, it combines the notion of geodesic distance to the well-known local outl...

This study demonstrates computational methods for the automatic detection and tracking of fish from video sequences. The research in this subject is very important especially in fish farming companies and for nature protection around the world. The process of automated control and counting of individual species of fish has a supportive contribution...

A new blockwise distortion measure is proposed to evaluate the quality of lossy compressed multispectral images. The measure is based on blockwise distortions which are calculated between the original multispectral image and the compressed/reconstructed multispectral image. The calculated measures with various compression ratios are matched to a vi...

A new area of machine learning research called deep learning has moved machine learning closer to one of its original goals: artificial intelligence and feature learning. Originally the key idea of training deep networks was to pretrain models in completely unsupervised way and then fine-tune the parameters for the task at hand using supervised lea...

The existing spectral colour difference metrics are not similar to CIEDE2000. The goal in this study was to implement a system to calculate the difference of spectral colours so that the calculated differences are similar to CIEDE2000 colour differences. The developed system is based on a priori calculated differences between known spectra and the...

Lossy compression of images corrupted by mixed Poisson and additive Gaussian noise is considered. Peculiarities of noise filtering effects observed for two approaches to image compression are studied. This is first done for specially created artificial image and, then, for two standard test images. It is shown that the standard (direct) approach to...

Light reflected from an object contains a range of information about its physical and chemical properties. Changes in the physical properties of an object can sometimes be evident as barely detectable changes of color. Our earlier study (Proceedings of the 15th Color Imaging Conference, Albuquerque, New Mexico, 2007. p 195–200) proposed a method fo...

The aims of this study are to develop the color density concept and to propose the color density based color difference formulas. The color density is defined using the metric coefficients that are based on the discrimination ellipses and the locations of the colors in the color space. The ellipse sets are the MacAdam ellipses in the CIE 1931 xy-ch...

In this study, modifying the CIE xyz color matching functions was considered to achieve a more uniform chromaticity space. New color matching functions resulting both from the non-negative tensor factorization and from the optimization were combined with two ellipse mapping approaches. In both approaches the original MacAdam ellipses were mapped to...

A michromatic (microscope plus chromatic) scope is a device that enhances the color discrimination between two spectral color datasets. Three spectral filters are required, instead of the conventional red, green, and blue filters, for the implementation of a michromatic camera. In this study, we describe two approaches to the design of these filter...

An improved method for the lossy compression of the AVIRIS hyperspectral images is proposed. It is automatic and presumes blind estimation of the noise standard deviation in component images, their scaling (normalization) and grouping. A 3D DCT based coder is then applied to each group to carry out both the spectral and the spatial decorrelation of...

In this paper we design and study an automatic approach to lossy compression of AVIRIS hyperspectral data. This approach takes into account the statistical characteristics of noise in component images as well as the considerable inter-channel correlation of data. An automatic method to estimate the noise variance in component images is proposed, an...

Digital imaging continues expansion to various applications. Spectral images are becoming more popular as one field of digital imaging. In this study we utilize a watermarking method for spectral images, based on the three-dimensional wavelet transform. We study the influence of watermarking process to illuminated watermarked images. In particular,...

Lossy compression of images corrupted by mixed Poisson and additive Gaussian noise is considered. Peculiarities of noise filtering effects observed for two approaches to image compression are studied. This is first done for specially created artificial image and, then, for two standard test images. It is shown that the standard (direct) approach to...

Lossy compression of images corrupted by Poisson noise is considered. Peculiarities on noisy images are discussed. An automatic approach in selection of a parameter that controls compression ratio (CR) is proposed. It is demonstrated how to reach optimal operation point (OOP), i.e., such CR that provides minimal peak signal-to noise ratio (PSNR) ca...

Computation of the non-negative tensor factorization of a spectral image is very time-consuming. The computational complexity depends on the number of bases, i.e. the rank of the factorization, and on the dimensions of the spectral image. In this study we propose sampling methods for the preprocessing phase which enables a faster way to compute the...

Communications in general require protection due to error-prone channels. In geoscience and remote sensing, especially coded or compressed data, and results from classifications are vulnerable to transmissions errors. Multiple descriptions of data are one way for protection of communications over unreliable channels. This study concentrates on mult...

We have studied the feature extraction from sea ice SAR images based on non-negative factorization methods. The methods reported here are the sparseness-constrained non-negative matrix factorization (SC-NMF) and Non-negative tensor factorization (NTF). The studies performed show that these methods can be used to extract meaningful features from SAR...

Two automatic approaches to lossy compression of hyperspectral AVIRIS images are proposed and considered. A first approach (strategy) is to filter images on-board and then to transfer compressed. A second strategy assumes that image filtering is performed on-land applied to decompressed data. In both cases, blind evaluation of noise variance is car...

In this paper we consider the use of XML technology for the data repository and data layer in the Hyperbook system. We propose the Hyperbook data layer (HBDL) for Hyperbook data repository using RDBMS. In HBDL the transformation technique from XML to relational data is the schema independent. The approach observes an XML file as consisting of terms...

The problem of selecting an appropriate wavelet filter is always present in the wavelet based image compression. In this study, we apply neural networks to wavelet filter selection. The purpose is to find a good filter for the com-pression of each multispectral image. Spectral character-istics from three different images are used in the training ph...

Hyperspectral images formed by AVIRIS imager are often compressed for data transmission and storage. For this purpose, both lossless and lossy compression methods can be used, but the latter are able to provide considerably larger compression ratios. Lossy compression can be done effectively in automatic manner with grouping of sub-bands and settin...

Volume clutter is three dimensional and is usually measured per volume. This clutter has large volume and includes rain, chaff, birds, and insects. Birds, insects, and other flying particles are often referred to as angel clutter of biological clutter. Chaff clutter is introduced as passive jamming. Chaff clutter is a kind of man-made distributed c...

Digital imaging continues its expansion into various applications. Spectral images are becoming more popular as one field of digital imaging. At the same time, the economic and security aspects in the utilization of images are finding increased emphasis. In this study, the authors apply a watermarking method for spectral images, whereby embedding i...

Digital imaging continues expansion to various applications. Spectral images are becoming more popular as one field of digital imaging. In this study we utilize a watermarking method for spectral images, based on the three- dimensional wavelet transform. We study the illumination influence to the watermarked images. The experiments were performed o...

The computation of non-negative tensor factorization may become very time-consuming when large datasets are used. This study shows how to accelerate NTF using multiresolution approach. The large dataset is preprocessed with an integer wavelet transform and NTF results from the low resolution dataset are utilized in the higher resolution dataset. T...

In this section we have collected experiences when different spectral images were compressed in a lossy manner with various methods described in the previous sections. The following abbreviations are used: • CL-W: wavelet transform in the spectral reduction followed by clustering, • CL-P: PCA in the spectral reduction followed by clustering, • WT-3...

The shape optimization of an elastic body in contact with a rigid surface is considered. An existence result for optimal shapes as well as a numerical realization are stated. From several numerical results it can be seen that minimizing the total potential energy of the system leads to an even distribution of contact forces on the contact boundary,...

Lossy compression of AVIRIS hyperspectral images is considered. An automatic approach to selection of compression parameters depending on noise characteristics in component images is proposed. Several ways of performing lossy compression are discussed and compared. It is shown that in order to minimize distortions and provide a sufficient compressi...

During the last few years the complexity of mobile phone services has increased notably due to rapid developments in lower level technologies. These developments are highly anticipated, but they have also affected the usability of both graphical and physical mobile phone user interfaces. In this paper we present information about strengths and weak...

A spectral filter is designed for enhancing the visual discrimination between two colors in a dataset. The filters are designed in spectral domain using spline interpolation through maximazing the color difference between the two selected colors. Two spectral datasets were applied in the design, the first dataset contained skin and vein colors from...

One type of volume clutters is rain clutter that is caused by rain droplets. Moving weather systems will have a nonzero Doppler response the rate at which the rain droplets approach to the radar system. The complete data radar collects contain the returns of both the target and the clutter. The signal processing block in a radar system uses filteri...

In this study we use non-negative matrix factorization (NMF) in deriving feature vectors from a set of spectral signatures. The purpose is to demonstrate the differences between the NMF and PCA feature vectors. The experiments show that NMF feature vectors are providing local features in spectral domain compared to the holistic features of PCA.

In this article we present new lossless compression methods by combining existing methods and compare them using AVIRIS images. These methods include the Self-Organizing Map (SOM), Principal Component Analysis (PCA), and the three-dimensional Wavelet Transform combined with traditional lossless encoding methods. The two-dimensional JPEG2000 and SPI...

In this study we apply the non-negative matrix factorization (NMF) to extract features from a spectral image. NMF features are sparse, they carry more localized information than the principal component analysis (PCA) features which are holistic in nature. The sparseness of the NMF feature vectors were controlled with a sparseness constraint. This a...

Moving weather systems will have a nonzero Doppler response at rate at which the rain droplets are approaching the radar system. The complete data the radar collects contain the returns of both the target and the clutter. The signal processing block in a radar system uses filtering operations to extract the target information while suppressing the...

The need to perform analysis and processing hyperspectral images registered by AVIRIS imager is stated. AVIRIS system basic characteristics are briefly discussed. Then, novel approaches and robust methods for estimation of noise variance and signal-to-noise ratios in band images are proposed. Using them, real life sets of AVIRIS images are analyzed...

In this work, a novel technique of objective spectral image quality evaluation is presented. The method is based on a Structural Similarity technique. The traditional approach, which deals primarily with gray-scale images, is extended to incorporate spectral data. The novel method has previously been tested against the conventional two-dimensional...

Companding is a procedure which can be used to smooth the effects of the fluctuating signal energy. The effect of the compressor in the transmitter is compensated with the expander in the receiver. PCA is an essential part in the lossy compression of the spectral images. In this paper we combine these two techniques to study the effect of compandin...

In this study we demonstrate the representation of sets of spectral signatures such that the representation is optimal in the user-defined sense. The basis functions for the sets are defined by minimizing the reconstruction error using the user-defined metric. In our proposal the basis functions are non-uniform rational B-splines (NURBS). NURBS rep...

In this study the purpose was to develop a computational surface model, which gives an equal computed chromaticity difference for equal perceived chromaticity difference in every part of the color space. The computational surface model assess chromaticity differences based on the surfaces defined by an ellipse data set and the two chromaticity poin...

A watermarking method for spectral images is proposed in this study. Multiwavelets are used in computing the transform domain of a spectral image. The visual, gray-scale watermark is transformed with the scalar wavelet transform and the transformed watermark is embedded in the three-dimensional transform domain of the spectral image. The strength o...

Companding is normally used in communication systems to compensate fluctuating levels in the source power. In this study we apply companding to enhance the coarse quantization of spectra. The spectra are selected from AVIRIS spectral images. Our experiments indicate that the signal-to-noise ratio (SNR) in coarser quantization of spectra may be enha...

Zero-order entropy has been applied as a basic scheme to the bit allocation among the eigenimages after the PCA transform in the lossy compression of spectral images. In this study, we enhance this allocation scheme by a nonlinear model. The exponentiation in the model is derived from optimization. The compression/reconstruction process is unsymmet...

In this study we present a technique to embed a digital watermark containing copyright information into a spectral image. The watermark is embedded in a transform domain of the spectral dimension of the image. The transform domain is derived by performing the principal component analysis (PCA) transform on the original image. The watermark is embed...

This study considers watermark embedding in spectral images. The embedding takes place in a transform space which is obtained through the Principal Component Analysis (PCA). The watermark is embedded in one eigenimage by mixing one eigenimage and the watermark. The watermark is a visual watermark which spreads to all bands of the image after the in...

The use of spectral images in many practical applications is getting more and more common as the cost of computation power and of storage space gets smaller. But like ordinary color images, the spectral images also suffer from noise caused by transmission and manipulation errors. In this study different multidimensional median filtering methods are...

Learning programming is a difficult task since programming requires new concepts in thinking and creative skills in problem solving. A number of learning tools and environments have been built to assist both teachers and students in introductory programming courses. In this study, we have established a classification for these tools. Tools are divi...

Digital imaging continues its triumphal march especially in geoscience and remote sensing. Digital information can be easily processed, manipulated and copied. Thus, the need for document ownership verification arises. In this study, we propose a watermarking method for spectral images. The watermark is embedded in a transform domain of the spectra...

We have developed a model to calculate chromaticity differences from surfaces defined from the MacAdam ellipses in CIE 1931 (x,y)-chromaticity diagram. The idea is to transform the MacAdam ellipses to circles on planes and to connect these planes to each other as two surfaces through interpolation. The distances along the surfaces are calculated by...

Environmental monitoring is an important task while estimating for example the influence of human activities in the nature. In this paper image analysis methods for performing these monitoring tasks are presented. We focus on the analyses of vegetation changes in lake water areas. Aerial photos taken in years 1996 and 1999 are used for the environm...

We propose a watermarking method for spectral images. The watermark is embedded in the three-dimensional wavelet domain of the image. Before embedding, the gray-scale watermark is transformed with the two-dimensional wavelet transform. After embedding the inverse three-dimensional wavelet transform reconstructs the image now containing the watermar...

This study proposes an interband version of the linear prediction approach for hyperspectral images. Linear prediction represents one of the best performing and most practical and general purpose lossless image compression techniques known today. The interband linear prediction method consists of two stages: predictive decorrelation producing resid...

Several powerful lossy compression methods have been developed for hyperspectral images. However, it is difficult to determine sufficient quality for reconstructed hyperspectral images. We have measured the information loss from the lossy compression with Signal-to-Noise-Ratio (SNR) and Peak-Signal-to-Noise-Ratio (PSNR). To get more illustrative er...

Several powerful lossy compression methods have been developed for hyperspectral images. However, it is difficult to determine sufficient quality for reconstructed hyperspectral images. We have measured the information loss from the lossy compression with Signal-to-Noise-Ratio (SNR) and Peak-Signal-to-Noise-Ratio (PSNR). To get more illustrative er...

Methods for noise reduction in multicomponent spectral images are developed and discussed. Multicomponent spectral images can be corrupted by noise either on all the channels or on some of the channels only. In the first case there are two possibilities: either the noise is on all the channels in the same way or the noise is randomly distributed on...

Automation of change detection greatly enhances environmental monitoring. This study describes possibilities for automation of change detection through image processing actions that include geometric correction, classification and segmentation. The water area was photographed twice and the changes in water vegetation were under consideration. The r...

In this paper we present a technique to embed a digital watermark containing copyright information into a spectral image. The watermark is embedded in the transform domain of the image. The transform domain is derived by performing the PCA transform on the original image. The watermark is embedded by modifying the coefficients of the eigenvectors f...

We have composed several lossy compression methods for multispectral images. These methods include the Self-Organizing Map (SOM), Principal Component Analysis (PCA) and the three-dimensional wavelet transform combined with traditional lossless coding methods. The two-dimensional DCT/JPEG, JPEG2000 and SPIHT compression methods were applied to eigen...

In this paper, a new group of noise reduction methods for multispectral images is presented. First, a 1-dimensional Self-Organizing Map (SOM) is taught using the pixel vectors of the noisy multispectral image. Then, a gray-level index image is formed containing the indexes of the SOM vectors. Several gray-level noise reduction methods are applied t...

In this report we examine the usability of different computational image segmentation and recognition methods as a tool of water area restoration projects. An airborne, mosaic image of Maavesi water area in South-east Finland was measured in September 2000 with a three-channel camera covering parts from the spectrum in red, green and infra-red wave...

Remote sensing produces large amounts of digital data that is collected into databases. Since a variety of applications utilize multispectral data, the data cannot be compressed with lossy methods for some user communities. In this paper, we propose improvements for the combination of two reversible methods for the lossless compression of multispec...

Remote sensing produces large amounts or digital data which are collected into databases. Since a variety of applications utilize the multispectral data, the data cannot be compressed with lossy methods. In this paper, we propose the combination of two reversible methods for the lossless compression of the multispectral images: first, principal com...

The problem of selecting an appropriate wavelet filter is always present in signal compression based on the wavelet transform. In this report, we give a method to select a wavelet filter for multispectral image compression. The wavelet filter selection is based on the Learning Vector Quantization (LVQ). In the training phase for the test images, th...

We have composed compression methods, which are based on the one-, two- and three-dimensional wavelet transforms. In addition, clustering the spectra of the image is considered. These compression methods, combining compression in the spectral and spatial domains, are compared by using a dataset of 65 multispectral images. The experimental results s...

Image compression has been one of the main research topics in the field of image processing for a long time. The research usually focuses on compressing images that are visible to humans. The images being compressed are usually gray-level images or RGB color images. Recent advances in technology, however, enable the authors to make the detailed pro...

The problem of selecting an appropriate wavelet filter is always present in the wavelet based compression. Different mother wavelets are characterized by their regularity, which describes the smoothness of the wavelet. Digital signals should be characterized similarly to enable the selection of a good wavelet filter. In this paper certain features...

We have developed three methods for the lossy com-pression of multispectral images. In this study, we com-pare the properties of those methods by applying them to several multispectral images. We also include results from several references. The experimental results show, that at low compression ratios, the best method is based on the wavelet trans...

Wavelet basics are presented. This is my own compilation of the wavelet theory. Contents 1 History 2 2 Continuous Wavelet Transform 3 2.1 Fourier Transforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2.2 Wavelet Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ....

Image compression has been one of the mainstream research topics in image processing. The research usually focuses on compressing images that are visible to humans. Images are usually gray-level images or RGB color images. Advances in technology enable one to make the detailed processing of spectral color features in the images. Therefore, compress...

In this paper, Daubechies orthogonal wavelets are used to compress color spectra. Color spectra have many forms from sharp peaks to constant value. Compression is needed to save space in storing and bandwidth in transmission of spectra. The underlying idea of this paper is to find suitable wavelets for compression. Quality of compression and recons...

Image compression has been one of the mainstream research topics in image processing. The compression methods are usually developed for images visible to humans, i.e. for gray-scale or RGB images. Recent advances in the field of remote sensing, however, require us to compress multispectral images. Many methods used in the traditional lossy image co...

In the first part we give a general existence theorem and a regularization method for an optimal control problem where the control is a domain in R**n and where the system is governed by a state relation which includes differential equations as well as inequalities. In the second part applications for optimal shape design problems governed by the D...

In relation to graphical user interface the physical user interface of mobile phones has evolved only on a detailed level during its life span. A large amount of research has been put to improve the usability of mobile phones' physical user interface and to respond to the increased complexity of mobile services. Still the physical user interface ha...

Learning programming skills is a demanding task for many first-year students. In this study we report results from Viope learning environment, which was used during one teaching period in August 2001. The results from the questionary and the exam show that Viope learning environment sup-ports the learning process and results to better grades in the...

In this report we describe the background and the properties of Aloha, which is a tool for learning structured programming. The learning approach in Aloha is based on understanding the program structures instead of the detailed syntax of the programming language. The implementation of Aloha supports both visual and active learning styles. The feedb...

Color differences are normally computed in the Lab color space with the CIEDE2000 color difference formula. As such, the color space is fixed and the difference formula is tuned for that space. This study is modifying the color space, starting from the Lab space such that the discrimination ellipses would reach both the similar shape and the simila...

In this paper, distance transforms are used in RGB color image compression. The compression is based on the idea of finding control points, e.g., points that are considered fundamental for the reconstruction of the image, from the three color planes. Calculation of these points is performed using the Distance Transform on Curved Space (DTOCS). In d...

Diss. -- Lappeenrannan teknillinen korkeakoulu.