Briefly describe the main technology of image authenticity check.

Image authenticity identification method

Based on the principle of image inspection, this paper analyzes and studies the common methods of image forgery, and introduces some methods to check the authenticity of images, such as photo quality inspection, photo repetition area inspection, consistency inspection of digital image and digital camera background noise, depth of field relationship of image content, consistency inspection of illumination direction, imaging perspective ratio inspection in photos, simulation photography inspection and object comparison inspection.

Keywords forged image inspection and recognition

The materials, tools and materials used in the process of forging photos exist objectively. At the same time, the composition, lighting, focusing, perspective relationship of various scenes, color correction and contrast control in post-production of photos all reflect the characteristics of whether the photos are complete or not, which provides feasibility for inspection and identification. Due to the rapid development of science and technology, it is difficult for computers to produce standards and data to identify computer forged photos. Based on the principle of photo recognition, this paper introduces the common methods of photo inspection.

1 test principle

1. 1 depth of field of photographic imaging

When taking pictures, we should focus on the subject, so that the subject can be clearly imaged on the focal plane. The scene in the scene space within a certain distance before and after the focal plane can also be a clear image for human eyes. People often refer to the spatial distance of the scene before and after the relatively clear imaging of the focused object as the depth of field. The phenomenon of depth of field is a relative concept, because the resolution angle of human eyes is the smallest. According to the principle of depth of field, the definition of objects in the depth of field in photos is inconsistent and shows a gradual trend. The closer to the focal plane, the higher the definition of the image. In addition, the foreground depth is smaller than the background depth of field. This imaging feature provides a theoretical basis for the identification of adding forged photos.

1.2 perspective principle of shooting and imaging

When the object is imaged on the focal plane through the optical lens, it conforms to the perspective principle of object imaging. Although the object is transformed from three-dimensional space to two-dimensional plane after being imaged in the photo, the positional relationship of the object in the photo still conforms to the perspective law in space. The perspective law of the object in the photo is as follows:

(1) Any straight line or plane parallel to the exchange surface remains unchanged on the screen and still maintains its original direction.

(2) Parallel straight lines of nonparallel graphs should disappear at one point, which is called vanishing point.

(3) The near is big and the far is small. It's because objects that are close to each other have a large angle of view, while objects that are far away have a small angle of view. Big viewing angle is big, small viewing angle is small.

(4) If the plane disappears into a straight line, this straight line is the elimination line. The vanishing line is the direction of the plane, and the difference of the vanishing line is the direction of the plane.

The perspective principle provides a basis for judging the height of people in photos and checking forged photos.

1.3 light consumption and illumination uniformity

Photography is imaging with light, which plays a role in both lighting and modeling. Different lighting angles and directions form different light lines and tones in photos. In photo inspection, the consistency of light lines and tones can judge the authenticity of photos.

When the lens is imaged, the unevenness of illumination on the phase plane determines the brightness of common color blocks in the image to change, that is, no matter how uniform the surface of the object is, there is no exactly the same block space in the photo. This principle can be used to test the cloned fake photos.

1.4 background noise of digital camera

The imaging elements (CCD or CMOS) of digital cameras usually have millions of photosensitive units. If one of the photosensitive cells is damaged and cannot be imaged, it becomes a bad pixel. Digital photography is different from traditional cameras, and the shooting environment of traditional cameras is much more complicated because of electronic parts. The residual energy on CCD is consistent with the fuselage parts from the temperature rise effect when the fuselage works, and even external electromagnetic interference may form mottled spots on the picture, which is noise. Bad spots and noise * * * constitute the background noise of a digital camera, which has a direct impact on the output of digital images taken by the camera. The originality of digital images can be tested by using the background noise of digital cameras.

Two kinds of computer forged photos

The original images of computer forged photos are mainly produced by digital camera shooting, scanner scanning and online downloading. Store external representations (size, shape, color, relative relationship, etc.). ) physical objects in digital form in memory. Because of the storage characteristics of images, digital images are easy to edit. The types of forged photos are:

(1) splicing forgery. By splicing different parts of different photos.

(2) Add forgery. Just add something to the photo screen.

(3) reduction and forgery. Is to cut or delete something on the photo screen.

(4) cloning and forgery. Using the clone seal in computer image processing to clone a part of the image, so as to achieve the purpose of forgery.

3 identification method

3. 1 General inspection of image quality

Routine observation and inspection are carried out on the printed or enlarged photos, combined with the possible characteristics of the spliced photos in the process of cutting, splicing and reproduction, such as image clarity, contrast size, color block size, color continuity, no normal patches, no image deformation, etc. , analyze whether the reason is left over from the cutting and splicing process. Observe the edge of the area where the photo may be forged with the magnification of live ammunition to see if there are any features such as wireless bar dislocation, false imaging, normal connection between images and no sleeve marks.

Only when the digital image needs to be enlarged to the pixel level can we observe the subtle changes of the image. At this point, we can find out the suspicious point or processing position of the image through the steps of gray level change and the color change of the color block. Then make the next specific confirmation.

3.2 Test affecting repeat region and clone region

Pre-check the local copy or regional query software with cloning technology, scan all areas of the same image, change the size of the scanned area, and find out similar blocks and duplicate blocks; For different images, find the similar or identical areas of the two images, and provide clues for further inspection.

By comparing and testing the images in the pre-exploration area, the similarities and differences are found out, and the causes, production methods and camouflage methods are analyzed, so as to identify the images accordingly.

3.3 Consistency Test of Digital Image and Camera Background Noise

For digital images, if there is a camera shooting images, we can use the background noise to identify the relationship between the image and the digital camera, which on the one hand verifies the relationship between the two, and on the other hand verifies the originality of the digital image, because the background noise of the image will change accordingly after image processing. The specific method is to use special digital camera background noise detection software to detect the background noise of the camera, get the background noise distribution of the digital camera, and then compare it with the noise points in the image to judge the consistency of the two.

3.4 Inspection of Depth of Field Relationship between Image Contents

For adding or cloning fake photos, the relationship between the sharpness of some parts of the image and the sharpness of the image between the foreground and the background will mostly violate the principle of depth of field. When checking an image, we can first find the focal plane of the image, and then check whether the change of object definition in the focal plane conforms to the principle of depth of field, so as to judge the authenticity.

3.5 lighting direction consistency check

The authenticity of the image is verified by the consistency of the lighting effect produced by using the light used in photography. The specific method is to analyze the direction, size, intensity and contrast of shadows formed by various scenes in the image in the irradiation direction of photographic light, find out suspicious points or differences, conduct numerical quantitative inspection and judge the reasons.

3.6 Inspection of imaging perspective ratio in photos

Based on the principles of geometric perspective and air perspective of photography, by comparing the size perspective, direction perspective, tone perspective and the proportional relationship between objects in the image, we can judge whether there is a phenomenon that violates the law in the image.

3.7 Simulated Photography Inspection and Physical Contrast Inspection

When there are suspicious points in the depth of field, illumination and perspective of the killed image, it is often necessary to verify the authenticity of the image through actual simulated shooting.

In addition, when checking the actual photographer, object and scene of the photo submitted for inspection, we can actually simulate the shooting according to the position relationship of the killed image through simulation photography, and directly check the coincidence degree and different positions of the person, object and scene in the photo through comparison to determine the objective authenticity of the relevant parts.

To sum up, the inspection of computer forged photos has both clear scientific theoretical basis and systematic inspection methods, but spear and shield are complementary opposites. With the development of science and technology, especially application software, the level of image forgery technology is constantly improving, which requires the joint efforts of technicians to explore and find more and more effective inspection means and methods.