Investigation of Binary Element Reproduction Methods in Elemental Record Processes Bohdana Havrysh1[0000-0003-3213-9747] , Oleksandr Tymchenko 2[0000-0001-6315-9375] Bohdan Kovalskyi1[0000-0002-5519-0759], Myroslava Dubnevych 1[0000-0001-7089-0190] Orest Khamula 1 [0000-0003-0926-9156], Michal Gregus 3[0000-0002-8156-8962] Mykola Logoyda 4[0000-0001-7597-7973] 1 Ukrainian Academy of Printing, Lviv, Ukraine dana.havrysh@gmail.com, bkovalskyy@ukr.net, dubnevychmyroslava@gmail.com, khamula@gmail.com 2 University of Warmia and Mazury, Olsztyn, Poland, o_tymch@ukr.net 3 Comenius University in Bratislava, Bratislava, Slovakia, Michal.Gregus@fm.uniba.sk 4 Lviv Polytechnic National University, Lviv, Ukraine, mykola.m.lohoida@lpnu.ua Abstract. We consider the methods of elemental recording of raster binary images on photographic or form material for systems of printing reproduction. In the process of tuning and comparing such processes, it is necessary to evalu- ate their sensitometric and structural-metric properties, which determine the quality of reproduction of the minimum image elements - raster points and strokes on the material carrier. The use of traditional sensitometry and structural analysis, which are used to calculate photographic processes, does not allow such a full assessment due to the difference between format and elemental re- cording methods, semitone and binary images, photographic and form materi- als. The peculiarities of the methods of reproduction of an image elements with elemental recording revealed to be expedient to be used at technological ad- justment of printing reproduction system processes. Keywords: binary images, optical density, printing forms, sampling, pixel functions. 1 Introduction and Problem Statement The basic working properties of the image elements on the material carrier are deter- mined by the gradation parameter. As a gradation parameter, optical density [1, 10] is often used to characterize the property of photographic forms to pass or delay radia- tion in the molding process, or the relative thickness of the printing layer. The register layer on the printing elements, which characterizes the property of printing forms to transfer ink in printing process [2]. For binary images, the gradation parameter has two levels - the upper lu and the lower ll, hich, depending on the polarity of the pro- Copyright © 2019 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0) 2019 DCSMart Workshop. cess, correspond to the active elements, purposefully formed radiation, and passive formed in the absence of radiation. If the active elements correspond to the lower level, then the process is positive (Fig. 1a), if to the upper one then the process is negative (Fig. 1b). The sensitometric curve (Fig. 1) shows the threshold exposure H thr to which the basic working property of the passive image elements is provided, and the minimum exposure H min from which the basic working property of the active elements is provided. If H min  H thr we get a threshold or step change in the grada- tion parameter. l 1.0 lu 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 ll lgH 0 0.15 0.30 0.45 0.60 0.75 0.90 1.05 1.20 Hthr Hmin a l 1.0 lu 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 ll lgH 0 0.15 0.30 0.45 0.60 0.75 0.90 1.05 1.20 Hthr Hmin b Fig. 1. Sensitometric curves of the elemental recording process: a - positive process; b - nega- tive process Sensitometric process curves will be used to identify the features of methods of re- producing image elements with elemental recording. 2 Development and Research of the Elemental Record Process Structure System The basis of the developed system of structurometry of the elemental recording pro- cess is based on the normalized functions of distribution of effective energy density: for the pixel - the function of pixel reproduction (FPR) and for the edges of the energy plate - the boundary functions in the direction of personnel (BFP) and line scan (BFL) [3, 11]. Reproduction functions are functions of a discrete argument. We use the following notations: P(y, x) – for FPR and E k  y  – for BFP. The x and y arguments related to the line and frame scans have a definition area [4]: 1 b : x, y  b, b  d , ,0, , b, where d  – sampling step N, b  N . N Kernel Energy Density - The fraction of energy density in the center of the image element from the maximum level of the energy plate obtained while recording pixels in all positions, and pixel kernel functions - the dependence of the kernel energy den- sity on the number of pixels forming active and passive elements, hka  k  and hkp  k  . For the purpose of accuracy of estimation, it is advisable to use a balanced raster structure [5], for which on the matrix of balance points with recorded pixels, active raster points are formed, and unwritten ones form passive points, with points in light and dark regions being formed by identical and round shape pixels. For the balance structure, formulas for calculating the energy density of the kernel for k-pixel image elements can be written using FPR: hka 1  P  0,0  , hka  2   2 P  0;0,5 etc. (1) hkp  k   hd  yc, yc   hka  k  where hd  yc, yc  – the level of the energy plate at a point corresponding to the center of the image element. Formulas (1) show that the energy density of the nucleus is completely determined by the central part of the FPR within the sweep step. 3 Binary Images Reproduction Research Reproducing extended image elements. Depending on the FPR, a constant unit level or periodic oscillations of the energy density level in the direction and with the step of the frame scan can be formed on the energy plate (Fig. 2). In this case, a maximum is formed in integer coordinates hdmax  1 (rationing to the maximum), and in half offset coordinates, – minium hплmin (Fig. 2, b). The average level of energy density on the plate is equal to that specified in the exposure system H w , then, in the transition from relative values to absolute values for the minimum and maximum, we can write: Rn hdmin and Rn H w , where 2 Rn  – is the conversion factor to the average level. In the presence of 1  hdmin oscillations it is necessary to distinguish two values for the threshold and minimum exposure - local and average. The first is related to the action of energy density in the maximum or minimum, the second - with the average value of the energy density on the plate and the exposure specified in the system. For offset printing forms, the main working property of the whitespace elements is provided in the complete absence of the printing layer  ll  0  , and printing elements at full thickness (lu  1) [6]. Then the main working property of passive elements with exposure increasing begins to break at the maximums in which the energy loc density operates at Hthr , with system exposure H thr : loc Hthr  Rn Hthr (2) The basic working property of the active elements will finally be reached in the loc lows in which the energy density operates at H min , at exposure H min : Rn R n max hef, rel. un. hef, rel. un. hds =1 max hef, rel. un. min 1 hef, rel. un. hds =1 min 1 min x x Rmin n hds E (y)x Ex (y) R nhhmin ds min ds E (y) E (y) hds yy y y a b Fig. 2. Edge functions in the direction of the frame scan: а – constant level of the plate; b – plate level fluctuations Exposure values H thr and H min can be determined experimentally, but values loc loc H thr and H min can be calculated by the formulas (2) and (3). With a constant unit   level of the energy plate hdmin  Rn  1 the exposure values set in the system are the same as the local values: Hthr  H loc thr and H min  H min loc . As the amplitude of the oscilla- tions increases Rn begins to increase and Rn hdmin decreases, then the threshold expo- loc sure of H thr is decreasing in relation to Hthr , and the minimum exposure of H min – is loc increasing in relation to the H min . Starting with H min the basic working properties of the long-lasting elements of the image will be provided, and this exposure can be considered to be minimally sufficient to reproduce them. Reproduction of small image elements. Image elements in the center of which the basic working property is provided are considered reproducible and form a pixel reproduction range of k а k p , where k а – is the number of recorded pixels that form the minimally reproduced active element of an image, k p – is the number of unwritten pixels that form the minimally reproduced passive image element. The pixel reproduction range is related to the reproduction range of tone values [7], which is one of the basic parameters that determine the visual   properties of printed product. The pixel width equals p  k а  k p  1 , where p – is the number of pixels in a raster cell, and increases with decrease of k а  k p . We will investigate the factors that affect this amount and width of the range. In the center of the k-pixel active image element the energy density acts like this: hkа  k  Rn H w , which for its reproduction should be not less then H min loc , then, consider- ing (3) we can write the reproduction condition: loc H min hmin H min hkа  k   hmin where hmin   d (4) Rn H w Hw In the center of k-pixel of the passive element, the image gives energy density h  k  Rn H w , which should be smaller then Hthr k p loc to recreate it, and considering (1) we can record the reproduction condition: hkp  k   hthr (5) loc H thr H where hthr   thr Rn H w H w Analysis (4) and (5) shows that starting with H min active and passive elements are reproduced, H w  H min – is the condition of the best reproduction of the passive ele- ments and the worst active, and the increase of H w comparatively to H min improves the reproduction of the active and the reproduction of the passive elements. hk ,rel. un. a hk(k) k =1 hmin h hthr p k =2 p hk(k) k Fig. 3. Graphs of pixel functions Let's consider the graphs of pixel functions for active hkа  k  and passive elements hkp  k  , which are piecewise linear, in one coordinate system (Fig. 3). The line hmin at the intersection with the curve hkа  k  will give us the point whose abscissa, when rounded, is greater than the number of pixels k а , that form the minimally reproduced active element. Similarly, the line hthr at the intersection with the curve hkp  k  gives us the point whose abscissa, when rounded to a greater number of pixels k p , that form the minimally reproduced passive element. Researching the influence of the values hthr and hmin , де hthr , hmin  0,1 and hthr  hmin , the sum of k а  k p allows us to draw the following conclusions.  The value of k а  k p depends on the width of the interval h  hmin  hthr and its position along the y-axis.  The smaller the interval h , the smaller the sum of k а  k p and the wider the pixel recreation range.  At a fixed value of the interval h , the minimum k а  k p corresponds ap- proximately to its central position on the segment of the y-axis, i.e. hmin  hthr  1 or H w  Hthr  hdmin H min . In this case, we obtain a symmetric pixel range with equal boundary values k а  k p . Some incertitude in the width of the range is introduced by the piecewise-linear nature of the func- tions, the angular coefficients of the linear sections of which depend on the FPR and the positions of the pixel recording.  The maximum possible pixel reproducing range can be obtained when hmin  hthr  0,5 , which corresponds to the threshold sensitometric curve  H min  Hthr  . The maximum possible range for a given raster structure is completely determined by the central part of the FPR. 1  Judging by (4) and (5) h ~ , the interval narrows with increasing ex- Hw posure in the system. However, as the exposure increases, the interval from the central position down the ordinate is shifted and k а  k p increas- es. As a result of these factors, in this case, the maximum range is asym- metrical with a smaller value of k а .  The piecewise linear nature of the pixel functions and the rounding of the pixels to integers results in the uncertainty of the bandwidth of up to two pixels. Also, the bandwidth is affected by the lower durability of the small printing elements not taken into account. Let's investigate the effect of the nature of the FPR distribution on the reproduction of small image elements. In the analytical method of calculating FPR as a factor con- trolling the distribution proposed analytical method of calculating FPR. As a factor controlling the distribution of the energy density of the FPR, we take r0 – the laser beam radius in the area of constriction at the level e 2 at constant scattering parame- ters, selected to approximate the experimental data obtained for offset heat-sensitive plates [8]. As r0 , decreases , the energy density fraction in the central part of the FPR increas- es, the values of the pixel function hkа  k  increase accordingly, and the values hkp  k    decrease (table 1), increasing the pixel range to full k а  k p  1 . The energy density values in the center of the 1-pixel active and passive elements are calculated by (1), and by (4) and (5), we calculate the conditions for their reproduction and calculate the exposures from which these conditions are satisfied: H loc loc H min hkа 1  min  H w1а  (6) Rn H w Rn L  0,0  loc loc H thr H thr hkp 1   H w1п  (7) Rn H w Rn 1  L  0,0   However, as r0 decreases, the amplitude of the oscillation level of the energy plate increases (columns hdmin and Rn of table 1), which can lead to the inadmissible parti- tioning of image elements into parts. The energy density at each point is formed by pixels that fall into the FPR definition area, and the minimals are formed in the coor- dinates of the frame sweep offset by half a step from the line item positions. Then the smallest value of the energy density is formed in the center of the 2-pixel element with the frame placement of pixels, for the calculation of which in formula (1) for the 2-pixel element with a row arrangement of pixels it is necessary to replace the argu- а ments of FPR: hmin  2  2L  0,5; 0 . Starting with r0  0,8 the energy density at the center of such an element begins to fall (Table 1). From (4) we can calculate the con- dition of the integrity of the elements of the image and calculate the exposure at which this condition is satisfied: H loc loc H min а hmin  2   min  H wint  (8) Rn H w 2 Rn L  0,5; 0  Table 1. Influence of laser beam radius on image element reproduction. H thr H min H w1а H wint H w1п hdmin Rn h 1 h 1 h    r0 a k k p а min  2 H thrloc loc H min loc H min loc H min loc H thr 1,20 1,00 1,00 0,37 0,63 0,53 1,00 1,00 2,72 1,89 1,58 1,10 0,99 1,00 0,42 0,58 0,57 1,00 1,00 2,36 1,73 1,73 1,00 0,98 1,01 0,49 0,51 0,62 0,99 1,01 2,02 1,60 1,94 0,90 0,94 1,03 0,57 0,43 0,64 0,97 1,03 1,71 1,51 2,25 0,80 0,87 1,07 0,65 0,35 0,64 0,94 1,07 1,43 1,46 2,70 0,70 0,75 1,14 0,74 0,26 0,59 0,87 1,17 1,18 1,48 3,35 0,60 0,57 1,27 0,81 0,19 0,48 0,79 1,37 0,97 1,63 4,20 0,50 0,37 1,46 0,87 0,13 0,32 0,68 1,87 0,78 2,11 5,28 4×4 1,00 1,00 0,97 0,03 0,98 1,00 1,00 1,02 1,03 33,11 It should be added that in the case of 4 × 4, a pixel consisting of 16 subpixels is used. With decreasing of r0 the exposure of H w1а decreasesfrom which the 1-pixel active element is reproduced, and the exposure of H w1p , to which the 1-pixel passive element is reproduced (the table shows the ratio of H w to local values). However, with r0  0,8 the minimum sufficient process exposure begins to be determined by the increased exposure of H wint integers required to maintain the integrity of the image elements (values highlighted in the table in bold). The need to increase exposure con- tradicts the lack of activity that is characteristic of, for example, the technology of elemental recording of offset printing forms. When using a pixel consisting of subpix- els of a larger recording extension [9], a constant single level of the energy plate is formed, and the full pixel range is reached without increasing the exposure (row 4 × 4 of the table). However, this approach requires specific hardware solutions. 4 Results and discussions As a result of the researches the following features of reproduction of image elements in the processes of elemental recording were revealed:  To obtain a larger pixel recreation range, the H thr threshold and the mini- mum H min exposure should be as close as possible to each other. The maximum possible range is reached in the case of a threshold sensitomet- ric curve  H min  Hthr  .  Choosing a working exposure H w in a system close to the minimum H min results in poor reproduction of the active elements.  The pixel range with equal lower and upper bounds is obtained at H w  Hthr  hdmin H min , but it is not maximal. The maximum pixel range is shifted toward larger exposures, with active elements reproducing better than passive ones. The result may be due to the lower resistance of small print elements to processing.  The integrity of the image elements can be controlled in the center of a 2- pixel pixel element.  Reducing the laser beam radius r0 allows us to obtain the full pixel range  k  k  1 , but this increases the oscillation amplitude of the energy а p plate and, as a consequence: increases the minimum H min , and decreases the threshold H thr of the exposure; the integrity of the image elements is impaired and the condition of its preservation determines the minimum sufficient exposure in the system. This increases the working exposure in the system, which may be unacceptable with insufficient activity, which is characteristic, for example, for the technology of elemental recording of offset printing forms.  The use of a pixel consisting of sub-pixels of greater recording expansion allows to obtain the full pixel range at a constant level of the power plate. It is expedient to use the revealed features of reproduction of image elements when designing and technological tuning of processes with elemental recording. 5 Conclusion Methods of elemental recording of bitmap raster images for photographic and form material for printing systems have been considered. The expediency of maximizing the minimum and maximum exposures with each other is shown, but the maximum pixel range is shifted toward larger exposures. To get the full pixel range, it is advisa- ble to use sub-pixels and reduce the laser pro-radius. References 1. Wang, H-A.: Data hiding techniques for printed binary images. In: Proceedings International Conference on Information Technology: Coding and Computing, Las Vegas, NV, USA, 55-59 (2001) 2. Lall, P., et al.: Effect of Process Parameters on the Long-Run Print Consistency and Mate- rial Properties of Additively Printed Electronics. In: 2019 IEEE 69th Electronic Compo- nents and Technology Conference (ECTC), Las Vegas, NV, USA, 1347-1358 (2009) doi: 10.1109/ECTC.2019.00209 3. Zahir, S., and Naqvi, M.: A near minimum sparse pattern coding based scheme for binary image compression. In: IEEE International Conference on Image Processing 2005, Geno- va, 285-289 (2005) doi: 10.1109/ICIP.2005.1530048. 4. Das, P., Paul, K.C., Chatterjee, A.: Optimizing luminance and chrominance filtered chan- nels by binary genetic algorithm (BGA) to generate color halftone image. In: 2015 Third International Conference on Image Information Processing (ICIIP), Waknaghat, 439-443 (2015) doi: 10.1109/ICIIP.2015.7414813. 5. Zhuge, X., Nakano, K.: An Error Diffusion Based Algorithm for Hiding an Image in Dis- tinct Two Images. In: 2008 International Conference on Computer Science and Software Engineering, Hubei, 684-687 (2008) doi: 10.1109/CSSE.2008.335. 6. Wu, M., Liu, B.: Data hiding in binary image for authentication and annotation. IEEE Transactions on Multimedia, vol. 6, no. 4, 528-538, (2004) doi: 10.1109/TMM.2004.830814. 7. Chen, Y., Hsia, C., Tsai, Y., Chi, K.: Halftone-Based Data Hiding Method Using Orienta- tion Modulation Strategy. In: 2018 1st International Cognitive Cities Conference (IC3), Okinawa, 192-194 (2018) 8. Shindo, Y., Kikuchi, N., Anada, K., Koka, S., Yaku, T.: Reduction of resolution for binary images by an octal grid graph representation model. In:2013 IEEE/ACIS 12th International Conference on Computer and Information Science (ICIS), Niigata, 417-422 (2013) doi: 10.1109/ICIS.2013.6607876. 9. Shu, J., Hu, A., Meng, F., Meng, Y.: Improved binary codes for efficient content-based image retrieval. In: 2015 8th International Congress on Image and Signal Processing (CISP), Shenyang, 550-554 (2015) doi: 10.1109/CISP.2015.7407940. 10. Dronyuk, I., Nazarkevych, M., & Poplavska, Z. Gabor filters generalization based on ateb- functions for information security. In International Conference on Man–Machine Interactions, 195-206, Springer, Cham (2017) 11. Peleshko, D., Rak, T., Peleshko, M., Izonin, I., Batyuk, D.: Two-frames image superreso- lution based on the aggregate divergence matrix. In: 2016 IEEE First International Confer- ence on Data Stream Mining & Processing (DSMP), Lviv, 235-238 (2016) doi: 10.1109/DSMP.2016.7583548.