Arto Kaarna

Compression of multispectral remote sensing images using clustering and spectral reduction

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 processing of spectral features in the images. Therefore, the compression of images with many spectral channels, called multispectral images, is required. Many methods used in traditional lossy image compression can be reused also in the compression of multispectral images. In this paper, a new combination of clustering spectra, manipulating spectral vectors, and encoding and decoding for multispectral images is presented. In the manipulation of the spectral vectors PCA, ICA, and wavelets are used. The approach is based on extracting relevant spectral information. Furthermore, some quantitative quality measures for multispectral images are presented.

An automatic approach to lossy compression of AVIRIS images

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 compression ratio it is reasonable to group the channels according to the evaluated noise variances in subband images and depending upon the sensor that produces sets of subband images. It is shown that for real life images the attained compression ratios can be of the order 8. ..25.

Linear prediction in lossless compression of hyperspectral images

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 residuals, and entropy coding of these residuals. Our method achieved an average compression ratio of 3.23 using 13 airborne visible/infrared imaging spectrometer (AVIRIS) images.

Digital watermarking of spectral images in PCA/wavelet-transform domain

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 spectral image. The transform is based on the principal component analysis and on the wavelet transform. With the inverse transforms the watermarked spectral image is reconstructed. The experiments indicate that the watermark is robust against attacks in lossy compression but more sensitive against mean filtering.

Classification of computerized learning tools for introductory programming courses: learning approach

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 divided into four categories: A) integrated development interface; B) visualization; C) virtual learning environments; and D) systems for submitting, managing, and testing of exercises. The classification is based on a review of existing tools, both commercial and freely available. Guidelines for the selection of a suitable tool are discussed.

Improved back end for integer PCA and wavelet transforms for lossless compression of multispectral images

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 multispectral images. Our improvements are three-fold: number of bits allocated to the coefficients from PCA is not constant but it is based on heuristics, difference between consecutive coefficients are entropy-coded, also the back-end is modified so that all bands are separately entropy coded, i.e. instead of one entropy coder we used several. Depending on the AVIRIS image, the actual compression ratios, calculated from the files sizes, were in the range from 3.05 to 3.21.

Multispectral image compression

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, compression of images with many spectral color channels, called multispectral images, is required. Many methods used in traditional lossy image compression can be reused also in the compression of multispectral images. In this paper a new combination of clustering of colors, manipulating spectral color encoding and decoding for multispectral images is presented. The approach is based on extracting relevant color information. Furthermore, some quantitative quality measures for multispectral images are presented.

Compression and classification methods for hyperspectral images

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 SPIHT compression methods were applied to the eigenimages produced by the PCA. The bit allocation for the compression of eigenimages was based on the amount of information in each eigenimage. In bit rate calculation we used the exponential entropy formula, which gave better results than the original linear version. The information loss from the compression was measured by the Signal-to-Noise Ratio (SNR) and Peak-Signal-to-Noise Ratio (PSNR). To get more illustrative and practical error measures, classification of spectra was performed using unsupervised K-means clustering combined with spectral matching. Spectral matching methods include Euclidean distance, Spectral Similarity Value (SSV), and Spectral Angle Mapper (SAM). We used two test images, which both were AVIRIS images with 224 bands and 512 lines in 614 columns. The PCA in the spectral dimension combined with JPEG2000 or SPIHT in the spatial dimension was the best method in terms of the image quality and compression speed.

Non-negative Matrix Factorization Features from Spectral Signatures of AVIRIS Images

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.

Multiwavelets in watermarking spectral images

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 of the watermark in embedding is controlled by perceptual constraints and the reconstruction quality. The experiments indicate, that the robustness of the proposed method against attacks like PCA/wavelet compression is better than in embedding in the transform domain obtained with the scalar wavelets