1 Assessment of paper surface topography and print mottling by texture analysis Aleš Hladnik * , Mirica Debeljak, Diana Gregor Svetec University of Ljubljana, Faculty of Natural Sciences and Engineering, Chair of Information and Graphic Technology, Ljubljana, Slovenia ABSTRACT Digital images of real objects are often characterized by regions of nonuniform intensities. Understanding image texture, defined as a function of the spatial variation in pixel intensities (gray values), is important in numerous applications. Texture patterns can be caused either by physical surface properties such as roughness, or they can be the result of reflectance differences such as the color or brightness of an object. In this paper two examples from papermaking and graphic arts industry are presented where various texture parameters – energy, contrast, correlation, inverse difference moment, entropy – were extracted from the gray level coocurrence matrices (GLCM). These were computed from the corresponding images acquired with laser profilometer (surface profile data) and flat-bed scanner (mottling) using ImageJ's GLCM Texture plugin. The results are discussed and compared with those obtained by conventional methods for determining paper surface topography and print mottling evaluation. Keywords: print quality, print mottling, paper roughness, image processing, texture analysis, gray level cooccurrence matrix 1. INTRODUCTION In papermaking and printing industries, precise characterization of topography of the paper substrate as well as assessment of its printed surface is of fundamental importance. For example, paper smoothness/roughness is one of the key parameters influencing the printing ink transfer in gravure printing technique 1 , paper and print gloss 2 , the quality of tissue papers and the friction of papers being fed into a printer without causing jams during printing. On the other hand, undesirable reflectance variation patterns often seen in offset printouts are related to uneven ink absorption and manifest as a cloudy, spotty print, known as print mottling 3 . With the rapid development and easy accessibility of powerful computers and image processing tools, it is now possible to visualize substrate's surface and quantify negative phenomena such as print nonuniformity – mottling – or excessive roughness of paper surface objectively, accurately and unambiguously. One of the many approaches used is also texture analysis, which is presented in this paper. 1.1 Texture Analysis Although surface properties of a material such as its smoothness/roughness, graininess, periodicity, homogeneity, directionality and others can be intuitively related to the corresponding digital image texture, it is extremely difficult to define this term formally. While Russ 4 characterized image texture as a descriptor of local brightness variation from pixel to pixel in a small neighborhood through an image, IEEE standard describes it as an attribute representing the spatial arrangement of the gray levels of the pixels in a digital image region 5 . Texture analysis has been extensively used in a variety of applications, such as medical imaging 6 , remote sensing 7 and paper inspection 8 and the implemented methods mainly differ in the way how textural features have been extracted. Apart from the structural, model-based and transform-based approaches 9 , one of the most frequently used statistical texture analysis methods is based on computation of the gray level coocurrence matrix (GLCM). GLCM – also known as the gray level spatial dependence matrix – is a matrix that keeps track of how often different combinations – pairs – of pixel intensity (gray level) values in a specific spatial relationship and distance occur in an image. From this matrix it is * Send correspondence to A. Hladnik. Further author information: A. Hladnik: E-mail: ales.hladnik@ntf.uni-lj.si, Telephone: +386 1 2003 280 M. Debeljak: E-mail: mirica.debeljak@ntf.uni-lj.si, Telephone: +386 1 2003 266 D. Gregor Svetec: E-mail: diana.gregor@ntf.uni-lj.si, Telephone: +386 1 2003 272