VideoCleaner Users Guide
VideoCleaner Users Guide
VideoCleaner Users Guide
7)
VideoCleaner can natively open a wide range of image and video files, including some that normally
require a proprietary player. For all other situations, VideoCleaner includes an optional free screen
capture utility (CamStudio) that can dub videos as they playback within their proprietary player.
ABOUT:
VideoCleaner is the world's most relied upon forensic video enhancement software because it:
Meets the highest evidentiary scientific standards validated by courts and experts worldwide.
Automatically maintains an audit trail of all filters and settings, and generates a metadata log.
Preserves evidence by applying industry accepted filters without altering the original recording.
Opens its source code for peer review scrutiny and is supported by hundreds of programmers.
Provides free unlimited updates and support, and is licensed for any legal purpose.
Is cost-free, ad-free, allows unlimited usage, and never collects any personal information.
With VideoCleaner, you can recover license plates and facial details, illuminate poorly lit scenes,
increase detail clarity, correct the viewing perspective, reverse lens distortion, repair VHS recordings,
improve color contrast, isolate channels, and so much more. VideoCleaner is cost and advertising
free, and created under the belief that clear evidence should be available to all those who need it.
Page 1 of 27
TABLE OF CONTENTS
Topic Page
Page 2 of 27
QUICK START PAGE
Launch "VideoCleaner" using the desktop icon. You will have
the option to Reload the default slider settings. The "No"
option is useful when working with multiple similar videos.
Adjust the program sliders to apply filters and features. The check boxes at the top of VideoCleaner
toggle on/off additional filter/feature sections. Once optimal clarity has been achieved, press the
computer's F7 key to save the video file (or use CTRL and S keys to save the current frame).
---------------------------------------------------------------------------------------------------------------------------------------
NOTES:
If the user activates conflicting VideoCleaner settings, or their computer's configuration prevents
a specific feature from working, an informational on-screen message will be displayed.
The program shortcut, “VideoCleaner fix”, resets VideoCleaner back to its installation defaults.
The program shortcut, “Reload last session” does just that, with all prior settings intact.
For technical support or feature requests, visit http://www.linkedin.com/groups?gid=2461639
KNOWN ISSUES:
VideoCleaner’s playback controls cause a stuttering view if your computer cannot keep up with the
filter processing demands. Use the right/left arrow keys or navigation bar to move to a specific frame.
Sometimes, pressing the "F5" (or ctrl-O) key will cause multiple file “open” windows to appear. This
occurs when Windows thinks that it hasn't yet offered the VideoCleaner file open selection window. If
this occurs, use the "Cancel" button (or the "ESC" key) to close the extraneous file selection windows.
If you are unable to see the sliders, even though the right-hand side panel is open, right-click on the
C:\VideoCleaner folder and select "Properties", then select "Edit" on the "Security" tab. Choose your
login, check the "Allow Full Control" option, and select OK to allow VideoCleaner to access its folder.
Page 3 of 27
USAGE HINTS
Looking for a fast image/video enhancement or authentication solution? Here are a few examples,
along with their corresponding VideoCleaner solution (denoted as feature group - specific feature).
Page 4 of 27
VERSION HISTORY (beginning with version 5.0)
5.7 Color limiter added to Forensic-RGBamplifier. DeGhosting added to VHS in Specialty section.
Negative Tools-Sharpening values now soften the scene.
5.6 New variables (inserted using the F2 key) to include file and system information into TEXT
annotations. VideoCleaner now loads for users who never followed the installation instructions.
5.5 Reorganized TOOLS “Deblur Denoise” slider, and added dirt denoise option. New CURVE “Data
bit depth” slider posterizes. Combined FORENSIC “Sobel” and “Canny” sliders. Swapped order of
FORENSIC “Isolate” and “RGBamplifier” sliders. Removed redundant MARGIN active slider.
Fixed error when all feature groups are made inactive.
5.4 New TOOLS “Hue” slider improves detail perceptibly through color shifting. New Append position
in CORE appends unaltered original prior to the enhanced results. Removed a slider redundancy
by allowing a Duplicate Frame Removal strength of zero to also serve as the filter’s on/off switch.
5.3 Significantly improved VideoCleaner’s automatic proprietary video extractor. Combined RESIZE
“Preserve aspect” and “Snap to common” sliders. New COMPARE slider Matrix option zooms in
on rectangular area defined in ACCENT. New “-1” “mask” position of ACCENT’s Pixelation slider
confines enhancements to user defined oval or rectangle accent area (unless using COMPARE’s
Matrix). The VELA file is now named and retrieved from the same folder as the file being tested.
5.2 F7 key also saves videos from within VideoCleaner. New top slider OVERRIDE option overrides
protective user limits. Increased TIMESTAMP annotation size range. FRAME number annotation
auto-resizes based upon aspect ratio. TRANSPARENCY slider moved to top (rightmost position
“7” replaces the now removed Bypass option). COMPARE’s “Original” text overlay is less opaque.
Removed “mid-frame” option (was redundant to the frame range options within the CORE group).
5.1 Combined TOOLS slider pairs (Backlight & Histogram) and (Denoise & Deblur), plus FORENSIC
slider pair (Motion & iFrames). Added 8th Forensic edging mode that most discards colored pixels.
New “Reload last session” program shortcut (also useful if installation step #1 was ignored).
5.0 FINALIZING section defaults to active. SuperResolution can be applied to any resizing. ACCENT
feature group defaults to a rectangular or oval highlight. DEBLOCKING and EQUALIZER feature
groups have been simplified, defaulted to their optimal mode, and relocated into TOOLS group.
The DESEQUENCER feature group was relocated into the SPECIALTY section. Added “Nearest
neighbor” to the RESIZE group, and Looping to the CORE group. Sub-second accuracy added to
time stamp annotation. Untouched video and its cropped version can now be placed beside the
fully enhanced version. Sliders streamlined to improve speed. Improved wording & error handling.
Page 5 of 27
THE VIDEOCLEANER INTERFACE:
Do not use the menu at the very top of this program. It relates to the AvsPmod interface and
not the VideoCleaner features. Instead, use the controls located on the right-hand side.
Script code: No need to Preview window: Shows the Section toggles: Check a toggle
examine this area unless you video in its current state. Apply box to activate and display the
want to modify or understand filters to make changes. Auto- features of that section. The
the VideoCleaner source code. resizes to fit the entire video. program automatically reloads.
If the VideoCleaner sliders are not visible, click on these 3 mini arrows to
restore their visibility. You can adjust the size of the control windows by
moving the boundaries guides (see the blue arrows in the image above).
NOTE: Avoid using Playback controls because most computers lack the power to play an enhancing
video in real time. Instead, use the computer's arrow keys or navigation bar to select a specific frame.
Page 6 of 27
SAVING YOUR WORK
VideoCleaner provides several methods to save one’s work. The user can hold down their computer’s
CTRL key and touch the letter “S” to save the current enhanced video frame as a single image, press
the ALT and “S” keys together to save all video frames as a series of images, or press the “F7” key to
save their enhanced results as a new video. All VideoCleaner settings are automatically retained, and
reloaded for reuse unless the user chooses to “Reload the file and lose the current changes” (which
resets all settings and sliders back to their defaults) the next time VideoCleaner is started. The latest
VideoCleaner settings are automatically logged in a file ending in "VideoCleaner_settings.txt" which
is located in the video's file folder (and also in C:\VideoCleaner). This log file is a powerful resource to
ensure that VideoCleaner results can be replicated on demand, and easily presented to the court.
To save only a portion of the video, the user only needs to select the first and last frames by moving
to the beginning and ending frame of the desired range and then selecting the relevant icon (see the
red arrows above). Similarly, pressing the computer's Delete key will remove only the designated
frame range. If frames are accidentally deleted, pressing CTRL-Z (holding down the "ctrl" key and
touching the letter "Z") will undo the last change.
Page 7 of 27
CAMSTUDIO
If a view cannot be opened in VideoCleaner, CamStudio can record that video while it the proprietary
software displays it onto the computer screen, including sound, and then save the resulting video so
that it can be opened by VideoCleaner. A low-resource ad-free version of CamStudio is ready to
install (Choose "Install - Camstudio screen capture", located in the VideoCleaner program shortcuts
folder). Once installed, follow these usage instructions:
Once the user presses the CamStudio Stop button, they will be asked for a file name and saving
location for the newly created recording. The recommended name is "Screen.avi" since that matches
the default input name for VideoCleaner. Be aware that screen captured recordings, while typically
much larger than the originating video file, will not contain any additional video details.
Once the newly dubbed video has been saved, it can be opened directly in VideoCleaner without the
need to use the proprietary viewing software. VideoCleaner can then be used to remove duplicate
frames and restore details to create a more accurate and compact dubbed video version.
WARNING: Depending on your computer, the dubbed version may include blended frames (hybrids
of two sequential moments in time) that compromise file accuracy and VideoCleaner enhancements.
Page 8 of 27
MEDIAINFO
The first time a video is loaded, VideoCleaner may create several support files in the same folder as
that video. One of those files may have a name extension that ends in ".ffindex". If this file exists, it is
a temporary index file created by FFDSHOW (the tool that opens most VideoCleaner's videos) to
index frames, and speed up future load times for that same video. Another file has a name extension
ending in ".Mediainfo.txt" and contains video metadata and facts useful for your reports and analysis.
The Mediainfo file may include GPS coordinates, camera settings, capture date/time, the creating
software name/version, and other identifying information. This file can be opened with any text editor.
Below is a simple Mediainfo.txt file for example purposes. Abbreviated metadata can be viewed within
VideoCleaner by moving the "File information" slider (located in the "Analysis" filter group) completely
to the right.
We recommended that you do not update the MediaInfo software as the newer versions include
adware and other potentially undesirable characteristics.
Page 9 of 27
VIDEOCLEANER - SLIDERS AND SETTINGS
Within each VideoCleaner section, are groups of features. Each feature is controlled by sliders. To
the left of the slider is its description plus the default (inactive) value shown in brackets. To the right of
the slider's name is the actual slider control. The number on the left edge of the slider control is the
lowest allowed value. The highest allowed value is listed to the right side of the slider. The farthest
right number (in blue) is the current value of that slider. Mouse clicking on the blue current value will
reset the slider to its setting from when the current video was last loaded or reloaded.
This is a group of sliders and NOT the
MAIN SECTION program's "TOOLS" drop down menu.
Page 11 of 27
FORENSIC (iFrame can be slow) (only some colors were definable): 1=black, 2=maroon, 3=brown, 4=red,
5=darkgreen, 6=darkoliveGreen, 7=sienna, 8=chocolate, 9=forestgreen, 10=olivedrab,
11=darkgoldenrod, 12=goldenrod, 13=limegreen, 14=chartreuse, 15=yellowgreen, 16=gold,
17=midnightblue, 21=darkslateGray, 22=dimgray, 24=tomato, 25=seagreen, 27=darkkhaki,
Apply FORENSIC settings (0=no) 28=sandybrown, 29=springgreen, 33=darkblue, 34=indigo, 35=dark magenta, 36=deeppink,
A value of zero (0) disables all features in this 40=palevioletred, 41=lightseagreen, 42=cadetblue, 43=dark gray, 44=tan, 45=medium
springgreen, 46=mediumaquamarine, 47=lightgreen, 48=khaki, 49=mediumblue, 51=blue
filter group. A value of one (1) allows them. violet, 52=fuchsia, 54=slateblue, 55=medium purple, 56=orchid, 57=dodgerblue,
58=cornflowerblue, 60=plum, 61=darkturquoise, 62=turquoise, 63=lightblue, 64=white
Use this slider if the video slightly alternates Flip This Video (0=no)
up/down as you use the computer right-left A value of zero (0) leaves the video alone. A
arrow keys to advance through the video. value of one (1) flips the video horizontally.
This can be required when a video Codec
incorrectly determines the video's left edge.
Nudge Fields (Off / Left / Right)
A value of zero (0) leaves the video untouched. Rotate Video Angle (0)
A value of one (1) will shift the even numbered Any value above zero (0) rotates the video
fields, of the now de-interlaced frames, leftward clockwise by that value's number of degrees.
one-half pixel. A value of two (2) will shift the The video is cropped to retain the video's
even numbered fields rightward half a pixel. dimensions. A value of "180" degrees can be
used with the "Flip" filter to create a vertical flip.
NOTE: Sub-pixel nudging, when required,
improves final enhancement results, but its Add Rotation Borders (0=no)
interpolation can impact Photogrammetry and A value of zero (0) leaves the video alone. A
Videogrammetry measurement accuracies. value of one (1) adds borders so the rotated
video remains fully visible without rescaling.
Page 14 of 27
CROP RESIZE
Page 15 of 27
SPECIALTY FEATURES SECTION
DESEQUENCE VIEW Perspective changer
Page 17 of 27
Relative Size (0=smallest) Percent from_Top (0)
The larger the value, the larger the relative size The value defines the percentage from the top
of the text annotation. Depending on the size of edge to place the timestamp. The default is
the target video, the default value of zero (0) zero (0) the top edge of the video.
may make the text too small to be seen.
Percent_from_Left (50)
Percent From Top (50) This slider sets the timestamp’s percentage
Percentage from the video’s top edge to show distance from the left edge. The default is fifty
the annotation. Default of fifty (50) is midway. (50), which approximates midway. As the value
approaches 99, you will effectively hide the
Percent From Left (0) fractional seconds portion of the timestamp.
Percentage from the video’s left edge to show
the annotation. Default of zero (0) is left edge. Relative_Size (4=medium)
The larger the value, the larger the relative size
Text Color (0=white) of the text timestamp, relative to the size of the
Each value uses a different annotation font target video. The default value is four (4).
color and transparency to ensure the text is
visible over nearly any video colors. The Starting Hours (0)
default value of zero (0) shows white text. Starting Minutes (0)
Starting Seconds (0)
Angle (0=horizontal) These slider values determine what timestamp
A positive value rotates annotation clockwise, is to be displayed upon the first video frame (in
while a negative rotates it counter-clockwise. 24-hour format). The timestamp increments
upon subsequent frames at the video’s frame
rate (adjustable in the PROPERTIES filter
NUMBER Frame Numbers group within the CORE section).
Surveillance video is saved with destructive (lossy) compression, which results in compression error
artifacts, and is perceived as high-energy moving noise. The continual absence or inconsistency of
"noise", "motion vectors”, and "DCT" (Discrete Cosine Transform) high-energy at any portion of a
video is a strong indicator of post-production editing or subjective blurring at that portion of the scene.
PseudoColor, converts subtle pixel intensity differentials into extreme color shifts, making it easier to
detect content cloning and intensity clipping, which will appear as identical colors. VELA (Video Error
Level Analysis) depicts content cropping or alterations tampered content as inconsistent brightness
and/or color, when compared to other scene details of similar contrast. Contrast or luminosity skewing
(e.g. excessive darkness/brightness) is depicted as blue/red respectively by the Min/Max test.
Open your video using VideoCleaner and, without applying any sliders, press F6 to transfer the video
to VirtualDub, but don’t close VideoCleaner. In VirtualDub, press CTRL-P and select the XVID Codec,
and now to the right you will see a "Configure" button (select that and in the third box down labeled
"Target quantizer" enter the number "8", select OK to save your settings), and click OK to commit the
XVID Codec. Now press F7 to save the video. Save the video in with the same folder and name as
the file being tested, but add “.VELA.AVI” to the file name. VideoCleaner will guide you to do this if
you forget. Close VirtualDub once the file saving process is complete.
In VideoCleaner, move the "Apply ANALYSIS settings" slider to the right to activate this feature group
and then move the "Test" slider to its rightmost position to activate VELA testing. Be aware that any
saved VideoCleaner.avi file will be automatically deleted the next time the VideoCleaner program is
re-started. As such, be sure to save or review your VELA results before closing VideoCleaner.
NOTES: Lossy compression or a skilled manipulator can affect testing results. Test results should be treated as
compelling evidence and not as a definitive conclusion. Prior training and understanding of test results is expected.
VideoCleaner analysis tests are based upon prior peer reviewed and public domain research: NOISE (Niabot,
Djembayz, and Waithamai), MOTION (Laurent de Soras and pavelsx2), DCT (N. Ahmed, T. Natarajan, and K.R.
Rao), Pseudo Colors (Dschwen, and Aimcotest), VELA (kaʁstn, George Chernilevsky, Dr. Neal Krawetz, and Doug
Carner), and MinMax (Anil. K. Jain, Fundamentals of Digital Image Processing, Prentice Hall, 1989 pp.240).
Page 19 of 27
TUTORIAL (Enhancing sample files)
The "VideoCleaner" folder on the user's computer includes a sub-folder named "Samples" which has
several practice files. Each file is listed below along with the recommended enhancement steps to
apply. Remember that, video enhancement is somewhat subjective. The quality of your vision and
computer monitor will affect your perceived results. You are encouraged to experiment with different
VideoCleaner features and settings. As with all skills, practice improves performance.
For these tutorials, it is recommended that the user install the Xvid Codec and Lagareth Codec, both
of which are listed as links in the VideoCleaner program shortcut menu on their computer.
First, use the desktop icon to launch the VideoCleaner program. When VideoCleaner loads, the user
then presses their computer's "F5" key (or CTRL-O) to load the desired file to work with.
Dark.jpg
Open the image file named "Dark" (or "Dark.jpg" depending on the computer’s configuration). It is so
dark that it is almost impossible to see anything, so first activate the TOOLS filter group by moving
the slider named "Apply TOOLS settings" rightward.
This is a very dark scene, so move the “Backlight / Off / Histogram” slider rightward to position 1
(histogram) to expand the lighting and color levels. Since the text appears faint, move the Video
contrast Strength slider rightward to about "14" to emphasize the luminosity differentials between
adjacent pixels (dots of resolution). Now you can add additional improvements by moving the
UnSharpen or Deblur sliders all the way rightward to improve image clarity.
NOTE: The Deblur filter can take minutes or even hours to process a video (which is why you should
always test it without applying any other filters when working with high-resolution videos. Since
"Dark.jpg" is just an image (a one frame video), the Deblur results will apply nearly instantaneously.
Turn ON the FORENSIC filter group (by moving Apply Forensic settings slider to the right) and set
Sobel to position "2". This filter will automatically locate and emphasize edges. Since position 2 is for
Chroma, and the image is nearly black/white, the resulting image takes on a greenish hue. To remove
the color, you need only turn ON the CHANNELS filter group and move the Off/ Gray / Negative /
Both slider Rightward to position “1”. Although, the green version may appear clearer due to the way
that our eyes work.
Now let's save the finished image by holding the computer keyboard’s CTRL key down while then
pressing the letter "S". Alternatively, the user can press their computer keyboard's "F6" key to send
this image into VirtualDub, where it can be saved in a wider range of image (FILE --> EXPORT -->
IMAGE SEQUENCE) or video file formats.
Page 20 of 27
XVID.avi
The "XVID.avi" video demonstrates an extremely dark video from a stationary camera. For this video,
you will need to have previously installed the XVID Codec. If this was never done, there is an
installation shortcut in the VideoCleaner program shortcut folder to take care of that now.
Load the "XVID.avi" video, which is located in the “Samples” folder of “VideoCleaner” on your
computer's C drive. As you play or scroll (right arrow key of your computer keyboard) through this
video you will likely see nothing except darkness. We can fix this by turning ON the TOOLS filter
group (by moving the slider named "Apply TOOLS settings" rightward), and then move the
“Backlight / Off / Histogram” rightward to position 1 (histogram). Now scrolling through the video we
can see a license plate, but each frame is not very readable due to the extreme pixilation noise.
Since this noise is random, we can average the video frames to improve the signal-to-noise ratio. To
do this, turn ON the FORENSIC filter group (by moving the slider named "Apply FORENSIC
settings" rightward), and then move the "Frame Averaging Radius" slider to its rightmost position,
"50". Immediately the random noise is suppressed. Since this setting averages the prior and
successive 50 frames, viewing frame number 50 (midway into the video) will maximize the effect.
Applying additional VideoCleaner settings will further improve the end results. For example, just
turning setting the Deblock Model to “1” (located in the TOOLS filter group) will apply a subtle effect
that leads to a slightly clearer image.
You can also move the "Video Contrast Strength" slider, located in the TOOLS filter group, to a
position of "10" to further gain additional subtle improvements, and move "Color Contrast Strength"
to a position of "5" to give the image a more visually pleasing appearance. That is the power of
having access to such an extensive enhancement toolbox within in VideoCleaner.
Now let's save the finished result by holding down the CTRL key and then pressing the letter "S",
which lets you save the current image in the location and format of your choosing. Alternatively, press
the F6 key to send the video to VirtualDub. Within VirtualDub, press the "F7" key to save your
enhanced video. If you hold the CTRL and press the P key (or "Compression" option under "Video"
menu) before pressing F7 Save, you can select a compression codec that will reduce the file size.
There is a slider named Deactivate all enhancements located above the Tools filter group. Move
that slider to the right to see what the image looked like prior to the applied enhancements.
Alternatively, you can use the transparency slider (the second to last VideoCleaner slider) to blend
the before and after versions. You can also use the OFF / Side / Atop / Sweep slider located near
the top to compare the before and after results.
Page 21 of 27
Noise.avi
Open the "Noise" video file. We can see that there is a lot of distracting noise in the Red and Green
color channels. Before we turn off color channels, we will need to ensure that the luminosity levels will
remain balanced. Turn ON the TOOLS section (by moving the slider named "Apply TOOLS
settings" rightward) and “Backlight / Off / Histogram” rightward to position 1 (histogram).
Move down several filter groups to locate and turn ON the Channels filter group (Apply CHANNELS
settings) and within that group turn OFF (move the slider rightward) both the Red and Green
channels, leaving a blue channel image. In this same filter group, move the Off/ Gray / Negative /
Both slider Rightward to position “1”. The image should appear much clearer but you can clearly see
that it is upside down.
In the Properties section is the ANGLE filter group. Turn this group on (Move the Apply ANGLE
settings slider rightward) and then change the Rotate video angle to a setting of "180". The image
is right side up, but we can see that it includes a noise pattern from having been a scanned image.
Change the rotation value to "181" and notice how the resulting interpolation reduces that pattern
effect.
We can further improve the the signal-to-noise ratio by adjusting the image's resolution using the
RESIZE filter group. Turn ON that group (also in the Properties section) and activate Preserve
General Aspect Ratio and adjust the New Width to "800". Like the angle change, here too we are
using interpolation to suppress the distracting noise from the image details.
At the top of all the sliders, mouse click a check mark to activate the Toggle "Specialty" Section and
be patient while the new feature group loads. Turn ON the COLORize feature group and set Filter
strength to "12", Low tones strength to "9", Add/Subtract red to "4", Add/Subtract green to "-2",
Add/Subtract blue to "4". These settings help restore a more natural color to the resulting image.
We can use the ACCENT filter group to define an area for highlighting (set the Use an oval to "1",
Highlight inner to "9", Highlight outer remains at "0", percent from_top to "60", percent
from_bottom to "24", percent from_left to "33, and percent from_right to "37"). Use the TEXT filter
group to place the word "Target" on the image (set the Relative size to "6", Percent from top to "55",
Percent from left to "27", Text color to "4", and Angle to "-18"). The actual text word (defaults to
"Target") and is changed using your computer keyboard’s F2 key.
Optionally, go to the LENS Fish Eye Barrel Correction section and use the top slider to turn this
filter group ON. Now play with the settings to see their effect. You can also try the VIEW section to
change the perspective (set the Top left (x) to "2" and Top right (y) to "4", with all other settings
remaining at their default).
You may wish to experiment with additional effects to see if the final results can be further improved.
Page 22 of 27
DCT.jpg
In nearly all circumstances, videos and images are compressed in a lossy manner, meaning that
details are discarded and error artifacts are introduced, while the perceived visual accuracy is
generally preserved. Each successive resave introduces additional compression errors. As such,
post-production manipulations will show different error artifacts from the remaining untouched scene.
Open the "DCT" image file (located in the C:\VideoCleaner\Samples folder) to see one of our forensic
lab stations. The picture appears normal, but it has been manipulated. Turn ON the ANALYSIS filter
section (by moving the slider named "Apply ANALYSIS settings" rightward) and then move the
"Extract (off/Noise/DCT/Motion)" slider to position number 2 to use the DCT (Discrete Cosine
Transform) filter feature. DCT reduces the variable differential from the scene's energy level, and this
is how most image and video files are compressed. In VideoCleaner, the DCT filter only displays the
variable component, which is where most compression errors reside. As such, once the DCT filter is
applied, the resulting scene may appear strange and quite dark.
Turn ON the TOOLS section (by moving the slider named "Apply TOOLS settings" rightward) and
then move the "Equalizer Strength" slider to its rightmost position to enhance the brightness
differentials. Notice that the area above the three chairs is devoid of noise, as are the three desktop
monitors. These are the altered portions of the scene. If you would like a more pleasant view, set the
"Enhancement Transparency" (the second to last slider in VideoCleaner) to about “4” to blend the
before and after results. In practice, DCT is highly effective at detecting in-painting, which is the most
common type of video manipulation.
DCT.jpg (the suspicious scene you are given) The original unaltered image
The VideoCleaner DCT test results depicting the manipulations (wall picture and desktop monitors).
Page 23 of 27
REFERENCES
Legend
A. Anil. K. Jain, Fundamentals of Digital Image Processing, Prentice Hall, 1989
(isbn:0133325784)
B. Marcus Weise & Diana Weynand, How Video Works, 2nd edition, Focal Press, 2007
(isbn:0240809335)
C. Vlado Damjanovski, CCTV Networking and Digital Technology, 2nd edition, Elsevier
Butterworth-Heinemann, 2005 (isbn:0750678003)
D. Bernard Grob & Charles E. Herndon, Basic Television and Video Systems, 6th edition,
McGraw-Hill, 1999 (isbn:0071163093)
E. Nosratinia, Enhancement of JPEG-Compressed images by re-application of JPEG, Journal of
VLSI Signal Processing, vol. 27, 2007 (issn:0922-5773)
F. J. Canny, A Computational Approach to Edge Detection, in IEEE Transactions on Pattern
Analysis and Machine Intelligence, vol. 8, no. 6, 1986 (issn: 0162-8828)
G. S.Murali, Govindraj B. Chittapur, Prabhakara H.S, & Basavaraj S. Anami, IJCSA vol. 2, no.6,
2012 "Comparison And Analysis of Photo Image Forgery Detection Techniques"
H. Advances in Visual Computing: 7th International Symposium ISVC 2011 Las Vegas, NV "A
Non-intrusive Method for Copy-Move Forgery Detection"
I. Hitesh C Patel and Mohit M Patel, IJCA (0975-8887) Volume 111, No. 15, 2015
"An Improvement of Forgery Video Detection Technique using Error Level Analysis"
J. W. Wang, J. Dong, and T. Tan, Digital Watermarking: 9th International Workshop, IWDW 2010
Seoul, Korea Section 3: "Tampered Region Localization of Digital Color Images"
*The development team of VideoCleaner is not compensated in any manner for the inclusion of filters
that have a demo license. These filters do not have any advertising or tracking, are fully functional in
their demo mode, and are included solely because they deliver superior peer-reviewed results.
Page 25 of 27
VIDEOCLEANER FILTERS ARE ALWAYS APPLIED IN THIS ORDER
(regardless of the order in which the user activates the program’s sliders)
DUPLICATE, CORE, FIELDS, ANGLE, CROP, RESIZE, CHANNELS (except negative), TOOLS +
FORENSIC + ANALYSIS, CURVE, negative, DESEQuencer, VHS, VIEW, COLORIZE, MARGIN,
ACCENT, TEXT, NUMBER, TIMEstamp, FINAL, subtract, COMPARE
NOTE: There is no such thing as a perfect order to apply filters that will always work best for all videos. The order
of the VideoCleaner filters are based upon the optimized results spanning thousands of real world forensic videos.
PRIVACY
VideoCleaner is intentionally designed to operate cleanly and anonymously on the most secure
networks without causing any exposure to security threats. VideoCleaner does NOT
communicate over the Network or Internet, does NOT collect any usage information, and does
NOT include any ad-ware, beggarware, etc.. of any type. VideoCleaner installs into one folder,
does not invoke any background or scheduled Windows™ tasks, and closing the VideoCleaner
interface immediately terminates all of its processing threads. The user is in full control at every
stage of installation and usage, and all source code is included for full operational transparency.
LEGAL
VideoCleaner can be redistributed and/or modified under the terms of each component's relevant
GNU General Public License, available in the VideoCleaner Support folder. VideoCleaner is
provided "as is" without any warranties or guarantees, and its installation or usage constitutes a
unilateral unlimited release of any and all related risks, damages and liability, actual or implied.
ACKNOWLEDGEMENT
Doug Carner donates his time as VideoCleaner’s lead programmer and support, and Forensic
Protection donates funding to keep the software and website advertising free. Neither party is
financially compensated, nor accept donations or pay, for any aspect of VideoCleaner, but gladly
accept accolades. Special thanks and appreciation are given to the invaluable development
contributions by Marc Robinson and Allen Combs. VideoCleaner would not be possible without
the peer review, user feedback and programming efforts of others. If you appreciate the features,
consider donating to the respective programmers (see the SCIENCE, MATH and CODE section
above). Users are encouraged to read the source code, create their own custom VideoCleaner
versions, participate in the free support forum, and communicate with the developers.
Page 26 of 27
GLOSSARY (definition of terms)
Codec - An abbreviation of the words Compression Decompression, and it defines (in computer
code) how a given video is to be saved and later reconstituted. Nearly all Codecs are lossy;
meaning that details are irreversibly discarded to reduce file size. These losses can appear as
muted brightness, an out-of-focus appearance, strange artifacts, and tiling (when the video looks
like a mosaic of tiny rectangles). VideoCleaner provides the option to save files uncompressed,
which preserves the available file details, thus creating very large file sizes. The noted defects can
be suppressed using VideoCleaner's Tools and Deblocking sliders.
Color space - There are different standards to record visual luminosity and color values. The
color wheel is based upon RGB (Red, Green, Blue) where each value is that color's contribution to
a given pixel. If all the RGB values are zero, the pixel appears black. If all the RGB values are at
their maximum, then the pixel appears white. Television, and thus nearly all security cameras, use
a planar color space, where pixel brightness is one value, and the color values are often shared
with their neighboring pixels. YV12 is the most common planar color space used with surveillance
video. The 2nd to last VideoCleaner slider transforms the enhanced video into YV12 color space.
Contrast - Linearly expand or contract the value difference between adjacent pixel. If the critical
details blend together, then VideoCleaner's video contrast (luminosity) and color contrast
adjustments can expand the subtle differences to make the visual details easier to recognize.
Adjust contrast with caution because, as contrast is increased, details located in the darkest or
brightest areas of the video can become clipped as they try to exceed the allowed values. Videos
that lack both very bright and very dark areas will benefit the most from contrast adjustments.
Histogram equalization - Non-destructively increases the global contrast of the entire video by
more evenly distributing the intensity differences between adjacent and overall pixel values.
Videos that are extremely bright (e.g. details are washed out) or extremely dark (e.g. a nighttime
video with insufficient lighting) will benefit the most from this process.
iFrames - Frames that contain the entire scene and thus are the truest to what the camera saw.
All other types of video frames are composites made by adding or subtracting pixels from the
closest iFrames. Depending on the video Codec, non-iFrames can introduce additional visual
defects. VideoCleaner's Forensic tools include an option to isolate just the iFrames.
Interlaced - When video broadcasting was invented (over 70 years ago), images from two
consecutive but independent moments in time (called fields) where simultaneously displayed as
one frame by interlacing these fields in a vertical weaving pattern. While newer systems use
progressive systems that avoid this (e.g. 1080i for interlaced and 1080p for progressive), most
security systems still record interlaced video. When those systems use their proprietary playing
software, that software will often deinterlace (undo the interlacing) by blending both moments in
time, which creates distortion and a false set of facts. VideoCleaner's Field filters (located in the
Properties section) can non-destructively isolate, and correctly align, those interlaced fields.
Pixel - A video or image is actually displayed as a grid of dots called pixels. Each pixel is denoted
by values that, depending on its format, may include details about luminosity (brightness), color
and transparency. VideoCleaner uses 32-bit RGBa color space to retain maximum pixel details,
and the status bar seen below the video displays the relevant values for the current image pixel
located under the computer's mouse pointer position.