IMAGE FORGERY DETECTION.ppt (Size: 2.5 MB / Downloads: 422)
IMAGE FORGERY DETECTION
• We are living in an age where we are exposed to remarkable array of visual imagery.
• Today's digital technology had begun to erode the integrity of images.
• Over the past few years ,the field of digital forensics has emerged to restore some trust to digital images.
• One solution to image authentication problem is digital watermarking.
• Water marking is actually a technique of message hiding in some image or text.
• The main advantage of using watermarks is to encode information which can prove ownership, e.g., copyrights .
• The drawback of this approach is that a watermark must be inserted at the time of recording, which would limit to specially equipped digital cameras.
Image Forensic Tools
• Over the last few years, there has been a growing body of work on tools for digital image forensics.
• These tools are capable of detecting tampering in images from any camera, without relying on watermarks or specialized hardware.
• Instead of watermarks, these tools assume that images possess certain regularities that are disturbed by tampering.
• The emphasis is on pixel
Techniques for detecting tampering:
• The most common image manipulations is to clone portions of the image to conceal a person or object into screen.
• If done with care, it is difficult to detect cloning visually.
• Two algorithms have been developed to detect cloned image regions:
PCA (Principal Component Analysis)
DCT (Discrete Cosine Transformation)
• To create a convincing composite, it is often necessary to resize, rotate or stretch portions of an image.
• Resampling introduces specific correlations between neighboring pixels.
• If it is known which pixels are correlated with their neighbors, then specific form of correlation is easily determined.
• But if neither is known, then a two step iterative algorithm EM (expectation/maximization) is used to solve the problem.
• A common form of photographic manipulation is the digital splicing of two or more images into a single composite.
• If spicing is done, then it will disrupt higher-order statistics which results in tampering.
• To randomly draw from the possible set of pixel combinations to obtain statistical regularities in natural images to detect image manipulations.
• Compute first and higher order statistics from a wavelet decomposition.
• Compute the bit agreements and disagreements across bit planes.
• Use linear search algorithm called Floating Forward algorithm, which differentiate authentic from manipulated images.
• Detect manipulations like resizing, filtering etc.
• Unique properties of lossy compression can be exploited for forensic analysis.
• Forensic technique that detect tampering deals with JPEG lossy compression scheme.
• RGB image is first converted into luminance/chrominance space .
• The full quantization is achieved by DCT method which produce a table of 192 values representing the channel.
• These values changes according to low and high compression rates.
• When an image block is individually transformed or quantized ,artifacts appear at the border of neighboring blocks.
• When an image is cropped or recompressed ,the specific pattern is disrupted.
• Only limitation is that there is some overlap across cameras of different makes and models.
• Both original and manipulated images are stored in JPEG format.
• In this scenario, the manipulated image is compressed twice.
• Double JPEG compression does not necessarily prove malicious tampering.
• Techniques for modeling and estimating different camera artifacts:
Color Filter Array
• Inconsistencies can be used as evidence of tampering.
• In an ideal imaging system, light passes through the lens and is focused to a single point on the sensor.
• Optical system, they fail to perfectly focus light of all wavelengths. The resulting effect is known as chromatic aberration.
• two forms: longitudinal and lateral.
• In both cases, chromatic aberration leads to various forms of color imperfections in the image.
• Lateral aberration manifests itself as a spatial shift in the locations where light of different wavelengths reach the sensor
• Local lateral aberration in a tampered region is inconsistent with global aberration.
Color Filter Array
Digital color image consist of 3 channels (RGB).
• Only a single color sample is recorded at each pixel location and other two samples are estimated from the neighboring samples.
• If specific form of correlation is known, then it is easy to determine which pixels are correlated with their neighbors.
• Any deviations from this pattern is an evidence for tampering.
• Most digital camera sensors are linear.
• There should be linear relationship between the amount of light measured by each sensor element and the corresponding pixel value.
• Difference in response function across an image results in tampering.
• When image is created by splicing together individual images ,there is often difficult to exactly match the lighting effects.
• There are techniques for estimating the properties of lighting environment.
• Differences in lighting across an image is the evidence of tampering.
• Today’s technology allows digital media to be altered and manipulated in ways that were simply impossible 20 years ago.
• As we continue to develop techniques for exposing photographic frauds, new techniques will be developed to make better fakes that are harder to detect.
• The field of image forensics, however, has made and will continue to make it harder and more time-consuming (but never impossible) to create a forgery that cannot be detected.