School of Electrical and Computer Engineering. Inkjet Forensics. Maria V. Ortiz Aravind K. Mikkilineni

Transcription

1 Inkjet Forensics Maria V. Ortiz Aravind K. Mikkilineni 1

2 About the Presenters Maria V. Ortiz received her B.S. from the Universidad Javeriana, Bogota-Colombia in She is a Graduate Student at Purdue University where she holds a Research Assistantship on the department of Electrical and Computer. Schlumberger awarded her a scholarship as a recognition of her academic effort in Her thesis, based in the recognition of the Malaria, was nominated to Best Engineering Thesis 2005 Aravind K. Mikkilineni - received his B.S. in Electrical and Computer Engineering from Ohio State University in Columbus, Ohio in In 2004 he received his M.S. in Electrical and Computer Engineering from Purdue University while working as a research assistant in the area of printer security and forensics. He is currently working toward a Ph.D. at Purdue University while continuing work on printer and device forensics. 2 2

3 Electronic Imaging Systems Laboratory Prof. Jan P. Allebach Investigate imaging systems (printers, scanners, digital cameras and displays) Characterize imaging systems Perform measurements and analysis on imaging systems Identify problematic sources in black box models Improve imaging systems Test solutions on systems Test solutions on human participants (psychophysics experiments) Improve solutions model to integrate the human perception Implement solution model 3 3

4 VIPER Lab Video and Image Processing Prof. Edward J. Delp Video Coding Device Forensics Watermarking Document Security Sensor Networks 4 4

5 Agenda Introduction to digital printing Basic inkjet printing process Digital Imaging Inkjet details Short Lab (Identifying print process parameters) Print modes (print driver settings) Examination of Print Samples [#] indicates a reference which are listed at the end of the presentation 5 5

6 Basic Inkjet Printing Process Paper path Carriage Cartridge Print head Inkjet printing technologies 6 The paper path along with the carriage, the cartridge, and the printhead are involved in every inkjet printing process and their interaction depends on the printing technology of the printer. 6

7 Overview of the Inkjet Process 7 7

8 Paper path [3] Paper is advanced through the printer by a series of rollers driven by a stepper motor. A carriage transports the pen back and forth across the page. The pen fires ink onto the surface of the page. 8 Red arrows show the printing direction Blue arrow shows the process direction 8

9 Carriage 9 These pictures were taken from the printer model we used in the demo session The region highlighted in red is an optical encoder strip which is used by the printer to position the carriage for proper dot placement. The image on the right shows the carriage which carries the printhead and the cartridge/cartridges. This printer uses two cartridges, one for color and the other for black. 9

10 Cartridge [2][9][21] 10 Different types of cartridges. The cartridges on the right do not carry the printhead. Instead, in this kind of printer, the printhead is fixed to the carriage in the printer. The middle cartridges have the printhead attached to them and as a consequence the printhead is replaced with the cartridge. On the left, we can see the electrical contacts between the cartridge and the printer. 10

11 Printhead/Nozzles 1/600 in 1/300 in 1/600 in 11 The printhead is composed of nozzles which eject drops of ink out of the cartridge and onto the paper. This diagram illustrates the nozzle placement for a printer with a resolution of 600dpi. Since the separation between them is very small, the nozzles are arranged in multiple offset columns (in this case two) because of mechanical limitations and to help avoid coalescence where two adjacent drops will merge into one drop. 11

12 Inkjet Printing Technologies [1] 12 Inkjet printing is divided into continuous and drop-on-demand technologies. In the continuous inkjet technology the drops are formed continuously but only some of them are expelled, on the other hand, in the drop-on-demand inkjet technology the drops are generated as needed. 12

13 Thermal Inkjet Side-shooter Roof-shooter [1] 13 The thermal inkjet technology is divided in two categories according to the position of the heater element, side shooter and roof shooter. As the temperature increases, the heater and the ink get hotter, until a bubble is formed and by the pressure in the ink reservoir, a drop is fired out the cartridge. 13

14 Bend mode School of Electrical and Computer Engineering Piezoelectric Inkjet Classification depends on the piezoceramic deformation mode. Four main types: Push mode Shear mode Squeeze-mode can be designed with a thin tube of piezoceramic surrounding a glass nozzle or with a piezoceramic tube cast in plastic that encloses the ink channel. Bend-mode design, the piezoceramic plates are bonded to the diaphragm forming an array of bilaminar electromechanical transducers used to eject the ink droplets. Push-mode design, as the piezoceramic rods expand, they push against ink to eject the droplets. Shear-mode the electric field is designed to be perpendicular to the polarization of the piezodriver. The shear action deforms the piezoplates against ink to eject the droplets. [1] 14 On the piezoelectric printers, the deformation of a piezoceramic material reduces the space inside the ink chamber and a drop is expelled. There are four categories depending on the piezoceramic element used. 14

15 Digital Imaging Image formation and associated artifacts 15 Understanding digital imaging artifacts and being able to distinguish them from printer artifacts is essential. 15

16 Digital versus Printer Artifacts Stepping: the imperfection of drawing of long lines sloped at a very small angle. [13] [19] 16 Stepping, the jaggedness seen on edges of straight or curved lines, is a digital artifact that could be mistaken for a printer artifact. 16

17 Halftoning 17 Understanding textures created by the process of halftoning are also important to understand. Small dots visible in light halftoned regions should not be mistaken for stray printer dots. 17

18 Inkjet Details Performance vs. image quality Ink drop formation and ejection Dot structure Swath alignment and Passes Multi-pass, single-pass, and print masks Page edge artifacts Paper interaction Ink types 18 18

19 Printhead Velocity Print Mechanism School of Electrical and Computer Engineering Performance Versus Image Quality Hardware Logic Pen size and Firing Frequency Output Resolution Ink/Media [21] Ink Capacity Ink Flux Ability to Mask Defects 19 Every process involved in creating an inkjet print is important and cause an inherent tradeoff between printing speed/performance and print quality. 19

20 Drop Ejection [14] 20 The drops are ejected out of the cartridge though the nozzles. As the drop falls from the cartridge to the media, it splits into the main body or primary drop and the secondary drop. The secondary drop should fall on top of the primary one but some aerodynamic effects prevent such event to happen. Notice that some of the ink is soaked up again by the empty space that was left in the ink reservoir. 20

21 Dot Structure Satellite Tail [3][4] 21 When the secondary drop does not fall on the same spot as the primary drop, it becomes either a satellite of a tail. Satellite : secondary dot is not attached to the main dot Tail : secondary dot is attached to the primary dot Using the position of the satellites or tails you can find out the printing direction Left to right: satellite/tail on the right side of the main dot Right to left: satellite/tail on the left side of the main dot leading the printing direction. 21

22 Effect of Carriage Speed on Dot Structure 15 ips left to right 45 ips left to right 45 ips right to left [3] 22 A slow print speed causes satellite drops to fall on top of or near the main drop as seen in the left image. As the print speed increases, the satellite drop falls further from the main drop as seen in the right two images. 22

23 Horizontal dot displacements for even raster School of Electrical and Computer Engineering Dot Placement Error Horizontal dot displacements for odd raster Vertical dot displacements for even raster Vertical dot displacements for odd raster [4] 23 Due to the position of the nozzles on the printhead (i.e. the space between the two columns) there is a misplacement of the dots in the horizontal position while in the vertical position there is not such error. 23

24 Swaths and Passes Swath height: length of printhead. Fixed size. Alignment: manual or automatic. Misalignment due to velocity and printing mode. [3][9] 24 Swath height: length of the printhead. In the diagram the size of the swath is 8, the first swath is printed from left to right and in the picture of the left you can see the spray/satellites on the edge, the second one is printed from right to left. The apparent misalignment from swath to swath is due to the high velocity of the carriage. 24

25 Single Pass Printing Horizontal Overlap Dark Swath Light Swath [21] Horizontal Gap 25 In the single pass mode the printer fires the nozzles one time per swath. The horizontal overlap/gaps are caused by an error in the advance of the paper and/or misdirected nozzles which are not firing perpendicularly and therefore such nozzles cover either more or less than the swath height. The dark and light swaths in a bidirectional printer are caused by the difference in the aerodynamics of each swath (i.e in one swath the carriage moves from L to R and on the next one moves from R to L). This aerodynamic difference makes the satellites fall closer/further in one direction than in the other. 25

26 Multipass Printing Print a given area in multiple passes Only some fraction of the dots in a swath are printed in each pass Minimize swath-to-swath errors Gaps/Overlaps Hue shifts Requires use of a print mask 26 Goals of the multipass mode 26

27 Print Masks Vertical position of pen for the 1 st pass Pen Sweep Direction Vertical position of pen for the 2 nd pass Media Advance Direction [3][4] 27 Example of two-pass printing mode. The swath is divided by the number of passes. In the first pass the printhead prints on the positions filled with ones in the print mask then the paper is advanced only half of the swath and the printhead uses the other half of the print mask to print. At the end of the second swath area to be printed in the rectangle will be finished. 27

28 Multipass Printing [21] 28 Another example of the multipass mode 28

29 4-pass Printing 1st Pass 2nd Pass 3rd Pass 4th Pass 29 The swath is divided by four (number of passes). In each pass only some spaces are filled and the paper is advanced one quarter of a swath height. At the end the top section will be finished. 29

30 Page Edge Defects Dot structure different during carriage acceleration and deceleration Accelerate Constant velocity Decelerate 30 The dot structure changes on the edges of the page because of the acceleration and deceleration of the carriage. The satellites/tails may look different from the ones present when the velocity is constant. 30

31 Paper-Ink Interaction [1][16] 31 31

32 Paper-Ink Interaction Paper samples 1 to 4 are inkjet coated papers, paper 5 is a polymeric laminate and paper 6 and 7 are coated copy papers. Paper 3 is the best coated paper and therefore its interaction with the ink is very similar to the laminate paper. [11] 32 32

33 Paper-Ink Interaction A noticeable defect with inkjet printing is wicking or feathering. This defect is caused when ink flows along or within cellulose fibers. Cockle effect: it happens when the paper gets wet from the ink [7][10][12][6] Difference in dot spreading between coated paper (left) and uncoated paper (right) 33 33

34 Ink Types [1] 34 34

35 Ink Types Dye-based ink Pigment-based ink 1 [1][15] Pigment-based ink

36 Demo Print the provided test page using each of N modes. Determine as much as you can about the print process parameters chosen by the driver for each of these modes. You should try to answer the following questions for each mode without physically examining the printer: Is the test page printed in single or bi-directional mode? Can you approximately guess the print speed by examining the page (fast or slow)? Was a multi-pass mode used? Is the page printed in only K, CMY, or a mix of both (CMYK)? What is the swath height? Are there different swath heights for K and CMY? How many columns of nozzles are there for K? Can you identify defects on the page due to the choice of these print process parameters? These questions are not exhaustive of the type of information that might be gathered from a printout, and it is possible that not all of the above questions can be answered for the test print in a given mode

37 Print Modes Print speed Number of passes Number of direction Drop volume Resolution 37 37

38 Example Draft Mode Fast print (carriage) speed Satellites and tails Single pass/bidirectional printing Visible swath boundaries Hue shifts Lower resolution Larger drop size 38 38

39 Example High Quality Mode Slow print (carriage) speed Minimize satellites and tails Multi-pass printing/single direction Reduce visibility of swath boundaries High resolution Small drop volume 39 39

40 Print Samples 40 Print samples (provided by Jan de Koeijer) 40

41 Print Sample 1 Bi-directional (Aerodynamics) More dot-gain in one direction Pen defects If thermal IJ heat buildup Difficult to tell with vignetting 41 Characteristics: bidirectional printer, probably single pass mode. Possible explanations: Assumption: swath size is one text line. There is more dot gain in one direction (even text lines) than in the other (odd text lines) because of the aerodynamics. The satellites sometimes fall very close or on top of the main drop, and sometimes far away. Assumption: swath size is two text lines. Half of the pen is damaged and therefore the nozzles are not printing properly. Assumption: temperature effects - the edge of the page from where the carriage starts the text line is lighter than the end of it because the temperature as well as the volume of ink is lower in that part. Difficult to tell because of vignetting 41

42 Print Sample 2 Swath discontinuity Aerodynamics combing magnitude difference 2 columns of nozzles 42 Characteristics: bidirectional printer. First segment printed from right to left because of the position of the combing, second segment printed on the opposite direction. Possible explanations: Assumption: swath size is the length of the second segment or longer. Draft mode. Because of the distance between the indentations it seems the pen has two columns of nozzles. There is more combing on one segment than in the other because of the aerodynamics of each direction. 42

43 Print Sample 3 Split in character Smart passing Not done in draft modes Ascenders/Descenders Nozzles used a lot fire well Nozzles not used a lot Run, more spray, offset dots Combing column spacing 2 column 43 Characteristics: splits in characters of the ascenders of the first and third line Possible explanations: Assumption: split occurs on the swath boundary. The misalignment between swaths and some misdirected nozzles could cause the splits. Assumption: nozzles used a lot fire well The nozzles of the ascenders are not used to fire (there are only two ascenders b and d ) as abundantly as the ones of the descenders, consequently these nozzles produce more spray and offset dots. Smart passing: on some printers the driver can decide to fit as much text as possible in one swath, in order to avoid the splits in misalignment in the characters, and to put the swath boundary in the blanks between the text lines (not used in draft mode) 43

44 Print Sample 4 Aerodynamic effects Nozzle firing history 44 Characteristics: this is not common. Possible explanations: Assumption: aerodynamic effects 44

45 Print Sample 5 Draft style High carriage speed Lots of spray 45 Characteristics: notice the presence of a lot of spray on the left edge of the characters. The printer direction: right to left. Possible explanations: Assumption: draft mode. The high velocity of the carriage in draft mode cause spray 45

46 Print Sample 6 1 st line bold? 2 nd line gray (not black)? 46 Characteristics: see slide 41 Possible explanations: Assumption: first line bold Assumption: second line could be gray instead of black on the document 46

47 Print Sample 7 Combing phase change could indicate swath boundary Dead nozzles or misdirected nozzles 47 Characteristics: change of phase in the combing and white horizontal lines inside the character. Possible explanations: Assumption: swath boundary The swath boundary could originate the change of phase on the combing (between the red lines) of the character. Assumption: misdirected or dead nozzles The white lines within the character are due to some damaged/misdirected or dead nozzles. The misdirected nozzles do not fire perpendicularly so they could lead to have white spaces. 47

48 Print Sample 8 Black and Color Black left-to-right Color right-to-left 48 Characteristics: black was printed from left to right and color from right to left. Possible explanations: Assumptions: 48

49 Print Sample 9 2-pass tails on both sides Possible bad pen Bad spray and/or tails 49 Characteristics: two pass printing mode (there are tails on both sides of the characters). Possible explanations: Assumptions: bad pen. 49

50 Print Sample 10 Smear Built up fibers, pet hair, etc. 50 Characteristics: the tails have different angles. Possible explanations: Assumptions: fibers, pet hair and any kind of dirt can be attached to the carriage. 50

51 Print Sample 11 Draft mode Piezo-electric Low firing frequency Larger drop size to cover same area Lots of spray 51 Characteristics: piezoelectric printer. Spray Possible explanations: Assumptions: draft mode Due to the low firing frequency of the piezoelectric printers, the dots need to be larger to cover the same area in a draft mode where a high carriage speed is used. 51

52 Print Sample 12 Draft mode Single pass Bi-directional Spray very far from main drops 1 st, 5 th and last line look like different type of text Print head temperature 52 52

53 Conclusions Inkjet printing is complex Artifacts from a specific printer will differ depending on the settings of the print driver, content printed, paper type, ink type, etc 53 53

54 References [1] Hue P. Le, Progress and Trends in Ink-jet Printing Technology, Journal of Imaging Science and Technology, Volume 42, Number 1, January/February 1998, pp [2] Jeff Tyson, How inkjet printers work, [3] Edgar Bernal, Improved pen alignment for bidirectional printers. [4] Osman Aslan, Gazi Ali et al, Print quality issues related to digital printing and forensic applications. [5] Eric Hanson, How an ink jet printer works, Hewlett Packard Laboratories. [6] Computer Friends Inc, Travel to the center of an inkjet cartridge. [7] Dave Brooks et al, Improvement of Ink Jet Printer Performance by Modifying Office Papers, Ink-Jet Components Division, Hewlett-Packard, San Diego, California. [8] Kenji Suzuki et al, Dynamics of Droplet Forming in Ink Jet Printer, Recent Progress in Ink Jet Technologies II, [9] Rob Beeson, Thermal Inkjet: Meeting the Applications Challenge, Hewlett Packard Company, Corvallis, Oregon. [10] Cheryl Katen, The Top 10 Breakthroughs in Thermal Ink Jet Technology, Hewlett-Packard Company, San Diego, CA. [11] Anne Mähönen et al, The Splashing of Ink Drops in CIJ Printing, VTT Information Technology Espoo, Finland

55 References [12] Makoto Torigoye, High-quality Imaging Technologies for Color Bubble Jet Printers, BJ Printing System Development Center, Canon Inc., Japan, [13] Ludwik Buczynski, Special Print Quality Problems of Ink Jet Printers, Warsaw University of Technology, Mechatronic Department, Warsaw, Poland, [14] Hongming Dong et al, An experimental study of drop-on-demand drop formation, Physics Of Fluids 18, 2006 [15] Joseph E. Johnson and James A. Belmont, Novel Black Pigment For Ink Jet Ink ApplicationsCabot Corporation, Billerica, Massachusetts. [16] Akira Asai et al, Impact of an Ink Drop on Paper Canon, Inc., Kanagawa, Japan. [17] Leonard Carreira et al, The Effect of Drying Rate on Inter-Color Bleed Xerox Corporation, Webster, New York. [18] Aidan Lavery and John Provost, Color-Media Interactions in Ink Jet Printing, Zeneca Specialties, Manchester, UK [19] [20] He-Jo Lee and Jan Allebach, Inkjet printer model based halftoning, IEEE transactions on image processing, vol 14, no. 5, May [21] Morgan Shramm, HP Lab University, June, Paris,

56 Questions/Comments? 56 56