عنوان مقاله
کمیت سنجی بردار چند متغیره جدید برای مقایسه موثر تصویربرداری فراطیفی
فهرست مطالب
مقدمه
استراتژی طراحی کتاب رمز
شیوه کوانتیزاسیون برداری چند متغیره
نتایج آزمایش و تحلیل
نتیجه گیری
بخشی از مقاله
طراحی کتاب رمز از طریق خوشه بندی کور (CBC)
بهترین روش برای طراحی کتاب رمز ، اجرای یک جستجوی جامع است که به تعیین مجموعه ای ساختارنیافته از کلمات رمز کمک می کند. از آنجایی که جستجوی کامل بسیار وقت گیر است، در نتیجه برای تسریع این فرایند و دستیابی به کتاب رمز ساختار یافته از جستجوی مقید استفاده می گردد. کاربردی ترین شیوه ها برای طراحی کتاب رمز عبارتند از: الگوریتمLBG ، کوانتیزاسیون برداری فازی (FVQ)، تولید سریع کتاب رمزKekre (KFCG) و روش مبتنی بر تبدیل کسینوسی گسسته (DCT) .
کلمات کلیدی:
Novel multivariate vector quantization for effective compression of hyperspectral imagery$ Xiaohui Li a , Jinchang Ren b,n , Chunhui Zhao a,n , Tong Qiao b , Stephen Marshall b a College of Information and Communication Engineering, Harbin Engineering University, Harbin, China b Dept. of Electronic and Electrical Engineering, University of Strathclyde, Glasgow, UK article info Article history: Received 31 March 2014 Received in revised form 30 June 2014 Accepted 3 July 2014 Available online 15 July 2014 Keywords: Hyperspectral imagery Fuzzy C-means clustering Image compression Multiple regression Remote sensing Vector quantization abstract Although hyperspectral imagery (HSI) has been successfully deployed in a wide range of applications, it suffers from extremely large data volumes for storage and transmission. Consequently, coding and compression is needed for effective data reduction whilst maintaining the image integrity. In this paper, a multivariate vector quantization (MVQ) approach is proposed for the compression of HSI, where the pixel spectra is considered as a linear combination of two codewords from the codebook, and the indexed maps and their corresponding coefficients are separately coded and compressed. A strategy is proposed for effective codebook design, using the fuzzy C-mean (FCM) to determine the optimal number of clusters of data and selected codewords for the codebook. Comprehensive experiments on several real datasets are used for performance assessment, including quantitative evaluations to measure the degree of data reduction and the distortion of reconstructed images. Our results have indicated that the proposed MVQ approach outperforms conventional VQ and several typical algorithms for effective compression of HSI, where the image quality measured using mean squared error (MSE) has been significantly improved even under the same level of compressed bitrate. & 2014 Elsevier B.V. All rights reserved. 1. Introduction Hyperspectral imagery (HSI), through capturing hundreds of narrow and contiguous spectral bands from a wide range of the electromagnetic spectrum, has great capability in deriving comprehensive details about the spectral and spatial information of the ground material. As a result, it has been widely used in many remote sensing applications such as agriculture [1], mineralogy [2] and military surveillance [3]. In HSI, improved image quality is always desirable for better processing, which in turn results in a trend for an increase in spatial/spectral resolution, radiometric precision and a wider spectral range. Consequently, the data volume in the 3-D hypercube increases dramatically, resulting in challenges for data transmission, storage, and processing. To reduce the volume of data, effective coding and compression become a natural choice in this context. Existing approaches for HSI compression can be divided into two main categories, i.e. lossless and lossy compression [4]. Lossless compression has been traditionally desired to preserve all the information contained in the image. However, the compression ratios which can be achieved with lossless techniques are limited. Lossless coding techniques include entropy coding and predictive modelling [5,6], where typical lossy compression approaches are transform based techniques [7,8] and vector quantization (VQ) [9,10]. In lossless compression such as predictive modelling, both intra-band spatial correlation and inter-band spectral correlation are used to determine a statistical model to estimate image values using partially observed data. The model and the estimation error are then encoded to represent the hypercube, where the performance relies on the correlation and statistical modelling [11,12]. Lossy compression yields higher compression ratio at the cost of introduced information loss. Despite the quality in the reconstructed image, these techniques are very popular, especially when the required compression could be achieved by lossy techniques. Moreover, the effect of the losses on specific applications in HSI have been assessed, such as target detection and data classification, showing that high compression ratio can be achieved with little impact in performance [7]. Several methods have been proposed for lossy compression of HSI, some of which are generalizations of existing 2D image or video algorithms, such as JPEG 2000 [13]. In [14], a Karhunen–Loeve transform was used to Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/optcom Optics Communications http://dx.doi.org/10.1016/j.optcom.2014.07.011 0030-4018/& 2014 Elsevier B.V. All rights reserved.
