Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

What is Natural Selection?01:32

What is Natural Selection?

129.8K
Natural selection is an evolutionary process in which individuals with survival-promoting traits reproduce at higher rates. These favorable traits become more common within a population or species. Naturally selected traits initially arise via random genetic mutations. In order for selection to occur, there must be variation within a population, the trait controlling the variation must be heritable, and there must be an evolutionary advantage for variation in the trait.
129.8K
Fixed Action Patterns01:06

Fixed Action Patterns

17.7K
A fixed action pattern (FAP) is a specific, hard-wired sequence of behaviors that occurs in response to an external stimulus, called a sign stimulus. The behavior is “fixed” because it is essentially unchangeable—proceeding similarly across individuals of a species every time it occurs.
17.7K
Solving Equations Graphically01:27

Solving Equations Graphically

558
Graphical methods provide an intuitive and visual means of solving equations by representing functions on the coordinate plane. These methods are especially helpful for estimating solutions, analyzing complex expressions, or understanding the behavior of functions.To solve an equation graphically, it must first be expressed in the form y = f(x). The solution to the original equation corresponds to the x-values where the graph intersects the x-axis, meaning where f(x) = 0.For example, the linear...
558
Graphical Representation of Inequalities01:28

Graphical Representation of Inequalities

225
The graph of the equation where y equals x squared forms a curve known as a parabola. This curve acts as a boundary in the coordinate plane, dividing it into distinct regions based on the relative position of points.When the equality sign in the equation is replaced with an inequality—such as greater than, less than, greater than or equal to, or less than or equal to—the graphical representation changes from a single curve into a broader shaded area that signifies the set of all...
225
Solving Inequalities Graphically01:24

Solving Inequalities Graphically

248
Solving inequalities graphically involves using a visual approach to determine where a mathematical expression meets a specific condition, such as being greater than or less than another value. By examining the position of a graph relative to the x-axis or another graph, it becomes possible to identify the range of x-values that satisfy the inequality. This method provides an intuitive understanding of solution intervals by showing where the inequality holds true.Graphical solutions to...
248
Nature and Nurture01:10

Nature and Nurture

22.4K
Many human characteristics, like height, are shaped by both nature—in other words, by our genes—and by nurture, or our environment. For example, chronic stress during childhood inhibits the production of growth hormones and consequently reduces bone growth and height. Scientists estimate that 70-90% of variation in height is due to genetic differences among individuals, and 10-30% of variation in height is due to differences in the environments that individuals experience,...
22.4K

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Extremely Long Spin Lifetime of Light-Emitting States in Quasi-2D Perovskites through Orbit-Orbit Interaction.

The journal of physical chemistry letters·2020
Same author

Organ-specific autoantibodies in Chinese patients newly diagnosed with type 1 diabetes mellitus.

Endocrine journal·2020
Same author

VS<sub>2</sub> and its doped composition: Catalytic depolymerization of alkali lignin for increased bio-oil production.

International journal of biological macromolecules·2020
Same author

Role of an autophagy/lysosome pathway in NF-κB pathway blocked pancreatic cancer Panc-1 cells.

International journal of clinical and experimental pathology·2020
Same author

Tuning the hierarchical nanostructure of hematite mesocrystals via collagen-templated biomineralization.

Journal of materials chemistry. B·2020
Same author

Synthesis of a New Water-Soluble Melatonin Derivative with Low Toxicity and a Strong Effect on Sleep Aid.

ACS omega·2020

Related Experiment Video

Updated: Feb 11, 2026

Deep Learning-Based Segmentation of Cryo-Electron Tomograms
10:25

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

10.9K

Distinguishing Computer-Generated Graphics from Natural Images Based on Sensor Pattern Noise and Deep Learning.

Ye Yao1,2, Weitong Hu3, Wei Zhang4

  • 1School of CyberSpace, Hangzhou Dianzi University, Hangzhou 310018, China. yyaoprivate@gmail.com.

Sensors (Basel, Switzerland)
|April 26, 2018
PubMed
Summary

This study introduces a deep learning method using sensor pattern noise and high-pass filters to accurately distinguish computer-generated graphics (CGs) from natural images (NIs). The novel approach achieves 100% accuracy, even with JPEG compression.

Keywords:
computer-generated graphicsconvolutional neural networkimage forensicsnatural imagessensor pattern noise

More Related Videos

Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

10.0K
Investigation into Deep Breathing through Measurement of Ventilatory Parameters and Observation of Breathing Patterns
08:34

Investigation into Deep Breathing through Measurement of Ventilatory Parameters and Observation of Breathing Patterns

Published on: September 16, 2019

12.1K

Related Experiment Videos

Last Updated: Feb 11, 2026

Deep Learning-Based Segmentation of Cryo-Electron Tomograms
10:25

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

10.9K
Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

10.0K
Investigation into Deep Breathing through Measurement of Ventilatory Parameters and Observation of Breathing Patterns
08:34

Investigation into Deep Breathing through Measurement of Ventilatory Parameters and Observation of Breathing Patterns

Published on: September 16, 2019

12.1K

Area of Science:

  • Digital Image Forensics
  • Computer Vision
  • Machine Learning

Background:

  • Photorealistic computer-generated graphics (CGs) are increasingly difficult to differentiate from natural images (NIs).
  • Existing methods for distinguishing CGs from NIs often lack robustness and accuracy.
  • The unique sensor pattern noise (SPN) imprinted by digital cameras offers a potential forensic artifact.

Purpose of the Study:

  • To develop a robust and accurate method for distinguishing CGs from NIs.
  • To leverage sensor pattern noise (SPN) and deep learning for CG detection.
  • To evaluate the effectiveness of high-pass filters (HPFs) in enhancing CG detection.

Main Methods:

  • Images (CGs and NIs) were preprocessed by clipping into patches.
  • Three high-pass filters (HPFs) were applied to remove low-frequency content and highlight SPN.
  • A five-layer convolutional neural network (CNN) was employed for patch classification, followed by a majority vote for full-image classification.

Main Results:

  • The proposed method utilizing three HPFs significantly outperformed methods with fewer or no HPFs.
  • The approach achieved 100% accuracy in distinguishing CGs from NIs.
  • High accuracy was maintained even when natural images were subjected to JPEG compression at a quality factor of 75.

Conclusions:

  • The integration of SPN, HPFs, and CNNs provides a highly effective solution for CG detection.
  • The method demonstrates superior performance compared to traditional techniques.
  • This technique offers a reliable tool for identifying synthetic media in digital forensics.