FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3012103007> ?p ?o ?g. }

• W3012103007 abstract "Archiving, restoration and analysis of damaged manuscripts have been largely increased in recent decades. Usually, these documents are physically degraded because of aging and improper handing. They also cannot be processed manually because a massive volume of these documents exist in libraries and archives around the world. Therefore, automatic methodologies are needed to preserve and to process their content. These documents are usually processed through their images. Degraded document image processing is a difficult task mainly because of the existing physical degradations. While it can be very difficult to accurately locate and remove such distortions, analyzing the severity and type(s) of these distortions is feasible. This analysis provides useful information on the type and severity of degradations with a number of applications. The main contributions of this thesis are to propose models for objectively assessing the physical condition of document images and to classify their degradations. In this thesis, three datasets of degraded document images along with the subjective ratings for each image are developed. In addition, three no-reference document image quality assessment (NR-DIQA) metrics are proposed for historical and medieval document images. It should be mentioned that degraded medieval document images are a subset of the historical document images and may contain both graphical and textual content. Finally, we propose a degradation classification model in order to identify common distortion types in old document images.Essentially, existing no reference image quality assessment (NR-IQA) metrics are not designed to assess physical document distortions. In the first contribution, we propose the first dataset of degraded document images along with the human opinion scores for each document image. This dataset is introduced to evaluate the quality of historical document images. We also propose an objective NR-DIQA metric based on the statistics of the mean subtracted contrast normalized (MSCN) coefficients computed from segmented layers of each document image. The segmentation into four layers of foreground and background is done based on an analysis of the log-Gabor filters. This segmentation is based on the assumption that the sensitivity of the human visual system (HVS) is different at the locations of text and non-text. Experimental results show that the proposed metric has comparable or better performance than the state-of-the-art metrics, while it has a moderate complexity.Degradation identification and quality assessment can complement each other to provide information on both type and severity of degradations in document images. Therefore, we introduced, in the second contribution, a multi-distortion historical document image database that can be used for the research on quality assessment of degraded documents as well as degradation classification. The developed dataset contains historical document images which are classified into four categories based on their distortion types, namely, paper translucency, stain, readers’ annotations, and worn holes. An efficient NR-DIQA metric is then proposed based on three sets of spatial and frequency image features extracted from two layers of text and non-text. In addition, these features are used to estimate the probability of the four aforementioned physical distortions for the first time in the literature. Both proposed quality assessment and degradation classification models deliver a very promising performance.Finally, we develop in the third contribution a dataset and a quality assessment metric for degraded medieval document (DMD) images. This type of degraded images contains both textual and pictorial information. The introduced DMD dataset is the first dataset in its category that also provides human ratings. Also, we propose a new no-reference metric in order to evaluate the quality of DMD images in the developed dataset. The proposed metric is based on the extraction of several statistical features from three layers of text, non-text, and graphics. The segmentation is based on color saliency with assumption that pictorial parts are colorful. It also follows HVS that gives different weights to each layer. The experimental results validate the effectiveness of the proposed NR-DIQA strategy for DMD images." @default.
• W3012103007 created "2020-03-23" @default.
• W3012103007 creator A5038952798 @default.
• W3012103007 date "2019-09-27" @default.
• W3012103007 modified "2023-09-26" @default.
• W3012103007 title "Subjective and objective quality assessment of ancient degraded documents" @default.
• W3012103007 cites W1527189241 @default.
• W3012103007 cites W1561538450 @default.
• W3012103007 cites W1569245971 @default.
• W3012103007 cites W1575884470 @default.
• W3012103007 cites W1576039173 @default.
• W3012103007 cites W1829569432 @default.
• W3012103007 cites W1964859077 @default.
• W3012103007 cites W1968590503 @default.
• W3012103007 cites W1971155098 @default.
• W3012103007 cites W1972006393 @default.
• W3012103007 cites W1975869473 @default.
• W3012103007 cites W1982471090 @default.
• W3012103007 cites W1984241378 @default.
• W3012103007 cites W1984516781 @default.
• W3012103007 cites W1987931295 @default.
• W3012103007 cites W1991317775 @default.
• W3012103007 cites W1996285419 @default.
• W3012103007 cites W1996682642 @default.
• W3012103007 cites W1998157976 @default.
• W3012103007 cites W1998187461 @default.
• W3012103007 cites W2002629090 @default.
• W3012103007 cites W2005446794 @default.
• W3012103007 cites W2008051965 @default.
• W3012103007 cites W2009272644 @default.
• W3012103007 cites W2017690106 @default.
• W3012103007 cites W2018918151 @default.
• W3012103007 cites W2019002170 @default.
• W3012103007 cites W2019353007 @default.
• W3012103007 cites W2026139930 @default.
• W3012103007 cites W2031291158 @default.
• W3012103007 cites W2031583378 @default.
• W3012103007 cites W2040763506 @default.
• W3012103007 cites W2042436931 @default.
• W3012103007 cites W2047269717 @default.
• W3012103007 cites W2054465923 @default.
• W3012103007 cites W2057979442 @default.
• W3012103007 cites W2062432103 @default.
• W3012103007 cites W2063360098 @default.
• W3012103007 cites W2083795040 @default.
• W3012103007 cites W2091471778 @default.
• W3012103007 cites W2092449865 @default.
• W3012103007 cites W2098311356 @default.
• W3012103007 cites W2102166818 @default.
• W3012103007 cites W2102397036 @default.
• W3012103007 cites W2107476778 @default.
• W3012103007 cites W2110365886 @default.
• W3012103007 cites W2111692787 @default.
• W3012103007 cites W2112215040 @default.
• W3012103007 cites W2118500430 @default.
• W3012103007 cites W2120038204 @default.
• W3012103007 cites W2120208461 @default.
• W3012103007 cites W2120504013 @default.
• W3012103007 cites W2120838001 @default.
• W3012103007 cites W2122787031 @default.
• W3012103007 cites W2126927985 @default.
• W3012103007 cites W2129644086 @default.
• W3012103007 cites W2133251102 @default.
• W3012103007 cites W2141983208 @default.
• W3012103007 cites W2142731874 @default.
• W3012103007 cites W2144764737 @default.
• W3012103007 cites W2146962865 @default.
• W3012103007 cites W2153582625 @default.
• W3012103007 cites W2154134281 @default.
• W3012103007 cites W2154759562 @default.
• W3012103007 cites W2156909104 @default.
• W3012103007 cites W2156928014 @default.
• W3012103007 cites W2159772167 @default.
• W3012103007 cites W2160244865 @default.
• W3012103007 cites W2161223164 @default.
• W3012103007 cites W2161907179 @default.
• W3012103007 cites W2162692770 @default.
• W3012103007 cites W2163370434 @default.
• W3012103007 cites W2167034998 @default.
• W3012103007 cites W2168023013 @default.
• W3012103007 cites W2202919542 @default.
• W3012103007 cites W2291445757 @default.
• W3012103007 cites W2294363656 @default.
• W3012103007 cites W2300642894 @default.
• W3012103007 cites W2500825094 @default.
• W3012103007 cites W2508672418 @default.
• W3012103007 cites W2551199247 @default.
• W3012103007 cites W2603172850 @default.
• W3012103007 cites W2605181485 @default.
• W3012103007 cites W2748431992 @default.
• W3012103007 cites W2767619925 @default.
• W3012103007 cites W2777280533 @default.
• W3012103007 cites W2963918210 @default.
• W3012103007 cites W615588766 @default.
• W3012103007 cites W79139011 @default.
• W3012103007 cites W1997781091 @default.
• W3012103007 cites W2017465979 @default.
• W3012103007 hasPublicationYear "2019" @default.
• W3012103007 type Work @default.
• W3012103007 sameAs 3012103007 @default.

• next