1、Designation: E2825 12Standard Guide forForensic Digital Image Processing1This standard is issued under the fixed designation E2825; the number immediately following the designation indicates the year oforiginal adoption or, in the case of revision, the year of last revision. A number in parentheses
2、indicates the year of last reapproval. Asuperscript epsilon () indicates an editorial change since the last revision or reapproval.1. Scope1.1 This guide provides digital image processing guidelinesto ensure the production of quality forensic imagery for use asevidence in a court of law.1.2 This gui
3、de briefly describes advantages, disadvantages,and potential limitations of each major process.2. Referenced Documents2.1 ISO/IEC Standard:2ISO/IEC 10918-1:1994 Information technologyDigitalcompression and coding of continuous-tone still images:Requirements and guidelines (JPEG) (also published asCC
4、ITT Recommendation T.81 (1992)2.2 SWGIT Material:3SWGDE/SWGIT Glossary SWGDE and SWGIT Digital and4.2.2 The end result is presented as a processed or workingcopy of the image.4.3 Avoid the introduction of artifacts that add misleadinginformation to the image or the loss of image detail that couldlea
5、d to an erroneous interpretation.5. Significance and Use5.1 Processed images are used for many purposes by theforensic science community. They can yield information notreadily apparent in the original image, which can assist anexpert in drawing a conclusion that might not otherwise bereached.5.2 Thi
6、s guide addresses image processing and related legalconsiderations in the following three categories:5.2.1 Image enhancement,5.2.2 Image restoration, and5.2.3 Image compression.6. Image Enhancement6.1 Image enhancement is any process intended to improvethe visual appearance of an image.6.1.1 Use bri
7、ghtness adjustment when the image is toobright or too dark. If the image is made too bright, there is arisk of loss of detail in light areas. If the image is made toodark, there is a risk of loss of detail in the dark areas.6.1.2 Use color processing to modify the color characteris-tics of objects w
8、ithin an image. This includes color spacetransformations, pseudocoloring, and hue and saturation ad-justments.6.1.2.1 Application of these techniques can compromise thecolor fidelity of the image.6.1.3 Use contrast adjustment when the image lacks suffi-cient contrast. If the image contrast is increa
9、sed too much,there is a risk of loss of detail in both light and dark areas.6.1.4 Use cropping to remove that portion of the image thatis outside the area of interest.6.1.5 Use dodging and burning to adjust brightness inlocalized areas.6.1.6 Use linear filtering techniques (see Fig. 1) to increaseth
10、e contrast of small detail in an image. These includesharpening, blur removal, edge enhancement, and deconvolu-tion. If a low degree of enhancement is used, the image willremain an accurate representation of the scene. If a high degreeof enhancement is used, the image may no longer be anaccurate rep
11、resentation of the overall scene, though it still maybe useful as an adjunct for interpretation of small details.FIG. 1 This Example Illustrates the Effects of Linear FilteringLeft: Original Image, Middle: Blurred Image, and Right: Sharpened ImageE2825 1226.1.6.1 A high degree of enhancement can als
12、o increase thevisibility of existing noise and artifacts; examples of noiseinclude film grain, snow appearing on a television screen, orrandom color dots.6.1.7 Use nonlinear contrast adjustments to adjust the con-trast in selected brightness ranges within the image. Theseinclude gamma correction, gr
13、ayscale transformation, and theuse of curves or look-up tables, or both.6.1.7.1 Anonlinear contrast adjustment can be used to bringout details in the shadow areas of an image without affectingthe highlight areas.6.1.7.2 A severe adjustment can cause loss of detail, colorreversal, or the introduction
14、 of artifacts, or a combinationthereof. (See Fig. 2.)6.1.8 Use pattern noise reduction filters to identify repeatingpatterns in an image and selectively remove them. This type offilter can be used to remove patterns such as fabric weaves,window screens, security patterns, and halftone dots.6.1.8.1 O
15、veruse of this technique will remove materialimage detail.6.1.9 Use random noise reduction techniques to reduce thecontrast of small detail in the image to suppress random noise.These include such filters as low-pass filtering, Gaussianblurring, median filtering, and speckle removing.6.1.9.1 Overuse
16、 of this technique will remove materialimage detail.6.1.10 Use warping to change the spatial relationshipsamong the objects in an image. It is analogous to printing aphotograph on a rubber sheet, then stretching the sheet indifferent directions, and then tacking it down. Warping can beused, for exam
17、ple, to remove perspective from an image or to9unroll9 a poster that was wrapped around a pole.6.1.10.1 Used improperly, warping can distort the naturalappearance of the objects in a scene.7. Image Restoration7.1 Image restoration is any process applied to an image thathas been degraded by a known c
18、ause (for example, defocus ormotion blur) to remove the effects of that degradation partiallyor totally.7.2 Information that has been totally lost in the imageduring the original imaging process cannot be replaced throughrestoration. However, partial restoration can be successful evenwhen total rest
19、oration is impossible.7.3 Restoration Techniques:7.3.1 Use blur removal to remove partially or completely animage blur imposed by a known cause.7.3.1.1 Blur removal differs from the image enhancementfiltering processes because the blur removal filter is designedspecifically for the process that blur
20、red the particular imageunder examination. Examples include defocus and motion blur,since these phenomena can be described mathematically. Thus,a specific filter can be designed to compensate for each blur.The degree to which a blur can be successfully removed islimited by noise in the image, the ac
21、curacy with which theactual blurring process can be described mathematically, andthe fact that information that has been totally lost cannot bereplaced. Often partial blur removal can be successful evenwhen total blur removal is impossible.FIG. 2 This Example Shows Nonlinear Contrast AdjustmentsLeft
22、: Original Image, Middle: Enhancement of Shadow and HighlightAreas at the Expense of Midrange Tones, and Right: Enhancement of Midrange Tones at the Expense of Shadow and Highlight AreasE2825 1237.3.2 Use color balancing to render the colors in the scenemore accurately. Color balancing is the extens
23、ion of grayscalelinearization to a color image and the adjustment of the colorcomponents of an image. For example, a color test targethaving known colors can be placed in the scene beforerecording the image. Then a grayscale transformation (nonlin-ear contrast stretch) can be designed for each color
24、 channel(red, green, and blue) to place the different colors on the testtarget in their proper relationship. This should reproduce theother objects in the scene in their proper relationship.7.3.2.1 Improper color balance can render colors inaccu-rately causing objects to appear to have the wrong col
25、or.7.3.3 Use grayscale linearization to render faithfully thedifferent brightness values in the scene. This adjusts thebrightness relationships among the objects in a scene. Forexample, a monochrome test target having known gray valuescan be placed in the scene before recording the image. Then agray
26、scale transformation (nonlinear contrast stretch) can bedesigned to place the different gray values on the test target intheir proper relationship. This should put the other objects inthe scene in their proper brightness relationship as well.7.3.3.1 Improper grayscale linearization can render bright
27、-ness values inaccurately so that objects may appear brighter ordarker than they actually appeared when the image wasrecorded.7.3.4 Use geometric restoration to restore the proper spatialrelationships among the objects in the scene. This restorationremoves geometric distortion from an image. It can
28、be used forthe removal of geometric distortion, such as that introduced bya curved mirror or a fish-eye lens.7.3.4.1 Geometric restoration differs from image warping inthat the geometric transformation is designed specifically forthe process that distorted the particular image under examina-tion.7.3
29、.4.2 The degree to which geometric distortion can besuccessfully restored is limited by the accuracy with which theactual distortion process can be described mathematically andthe fact that information that has been totally lost (for example,hidden behind another object or obscured from the camera)c
30、annot be replaced. Often, partial geometric restoration can besuccessful even when exact geometric restoration is impos-sible.8. Image Compression8.1 Digital images produce a large amount of data to bestored. Image compression techniques reduce the storagerequirements by making image data files smal
31、ler.8.2 Compression Processes:8.2.1 Lossless compression reduces file size by removingredundant information. Because the redundant information canbe retrieved to display the image, lossless compression resultsin no loss of information. Lossless compression does not alterthe content of an image when
32、it is decompressed.8.2.2 Lossy compression achieves greater reduction in filesize by removing both redundant information and data deemedexpendable by the compression algorithm. Because the ex-pendable data cannot be retrieved upon reconstruction of animage for display, compression results in some lo
33、ss of imagecontent as well as the introduction of artifacts.8.2.2.1 Degradation occurs each time the image is com-pressed using a lossy process, such as saving to a compressedformat.8.2.2.2 Higher compression ratios result in the loss of moreinformation. Normally, the degree of compression can bespe
34、cified.8.2.2.3 Depending upon the application, lossy compressionmay render an image less useful.8.2.3 The Joint Photographic Experts Group developed animage compression standard known as JPEG (ISO/IEC 10918-1:1994). This compression algorithm is applied to the image in8 by 8-pixel blocks. Normally,
35、it is used as a lossy compressionscheme in which the degree of compression can be specifiedbefore storing the image. However, JPEG can also be used asa lossless compression scheme. At high-compression ratios,JPEG could remove important image detail and introduceblocking artifacts as the block bounda
36、ries become visible (seeFig. 3). JPEG is but one of many compression algorithms.8.2.3.1 Compression should be used with care to avoidmaterial degradation of the image.8.2.3.2 The compression settings used by one camera orsoftware program may not be the same as the compressionsettings used by another
37、 camera or software program.8.3 Use of Compression:8.3.1 Many digital cameras store images using JPEG com-pression, so that some compression is unavoidable. Somedigital cameras are capable of storing images in an uncom-pressed form. The degree of compression should be set lowenough that material ima
38、ge content is not lost or obscured byartifacts.8.3.2 In instances in which the primary or original image isalready compressed, it should not be further compressed usinglossy compression processes; additional data will be lost.Sources of compressed primary images may include electronicbooking photogr
39、aphs, some types of digital camera images,and images downloaded from the internet or email. The fileformat is not an indicator of the compression history for animage. For example, a Tagged Image File Format (TIFF4) filemay have been previously compressed using a lossy algorithm(1).58.3.3 The materia
40、l use of an image may change over time.Any compression used to save an image should be appropriatefor the intended use at that time.8.3.3.1 Images intended for laboratory analysis should notbe compressed using a lossy process unless the resulting imagestill retains the relevant information as determ
41、ined by thelaboratory personnel conducting the analysis.9. Guidelines for Digital Image Processing StandardOperating Procedures9.1 The purpose of image-processing procedures is to applyprocessing techniques intended to enhance, restore, or com-press digital images, or a combination thereof. Standard
42、operating procedures should be developed and followed. A4TIFF is a trademark of Adobe Systems Incorporated.5The boldface numbers in parentheses refer to a list of references at the end ofthis standard.E2825 124sample standard operating procedure is included in the SWGITdocument “Guidelines for Image
43、 Processing” (2). See also theSWGDE/SWGIT document, “Recommended Guidelines forDeveloping Standard Operating Procedures” (3).9.2 EquipmentThe laboratory standard operating proce-dure (SOP) should define minimum hardware and softwareequipment requirements including, but not limited to:9.2.1 Hardware:
44、9.2.1.1 Input/capture device,9.2.1.2 Image-processing systems,9.2.1.3 Output devices, and9.2.1.4 Storage/archive.9.2.2 Software:9.2.2.1 Image management, and9.2.2.2 Image processing.9.3 ProceduresLaboratories should establish specificstep-by-step procedures for image processing according topublished
45、 guidelines. These procedures should address thefollowing as a minimum:9.3.1 Capture,9.3.2 Processing,9.3.3 Storage and archiving,9.3.4 Image management,9.3.5 Data security, and9.3.6 Output9.4 CalibrationLaboratories should develop SOPs forcalibrating all equipment that produces test results. Thesep
46、rocedures should be consistent with the manufacturers rec-ommendations.9.5 LimitationsLaboratories should document the limita-tions of their processes and equipment in their SOPs.9.6 SafetyLaboratories should develop safety proceduresspecific to their needs.9.7 ReferencesLaboratories should maintain
47、 their labora-tory specific documentation, manufacturers manuals, andpublished guidelines.9.8 TrainingLaboratories should define the level of train-ing necessary to perform the procedure. Refer to the SWGIT“Guidelines and Recommendations for Training in ImagingTechnology in the Criminal Justice Syst
48、em” (4) and “SWGDE/SWGIT Guidelines and Recommendations for Training inDigital and Multimedia Evidence” (5).10. Keywords10.1 criminal justice system; digital image processing; im-age processingFIG. 3 Left: Original Image, Middle: the Result of JPEG Compression (Compression Ratio = 15:1), and Right:
49、the Result of EdgeEnhancement after CompressionE2825 125APPENDIX(Nonmandatory Information)X1. SAMPLE STANDARD OPERATING PROCEDURES FOR LATENT PRINT DIGITAL IMAGING (LATENT PRINT UNITSLABORATORY DIVISION)X1.1 PurposeX1.1.1 This guide sets forth the Latent Print Units (LPU)specific procedures for latent print digital imaging.X1.2 Changes and ReviewX1.2.1 The section chief and unit chiefs are the only personswho may authorize changes to this guide.X1.2.2 The appropriate LPU personnel who handle evi-dence that may be digitally processed shall review the LPUstandard operatin
