A Method for Differentiating Physical and Non-physical Planar Images Using Spectral Analysis
A physical and image technology, applied in the field of image recognition, can solve the problems of non-reuse of information, high cost, narrow application field, etc., to achieve the effect of simple algorithm implementation, low storage pressure, and easy integration
Active Publication Date: 2022-02-25
上海泛洲信息科技有限公司
View PDF5 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
This kind of technology is complex and complex, and some technologies are still in the development and transition period, the cost is high, and the information is not reused, resulting in an excessive amount of information or transmission in the entire system; the application field is narrow
Especially when the imaging position does not move and the target object does not move, it will result in the inability to obtain different frames of images between multiple dynamic images, and it is easy to identify the real object as a non-real object
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0046] Below in conjunction with accompanying drawing and embodiment the present invention will be further described:
[0047] All current solutions basically focus on the time domain information of images, but the present invention breaks through the time domain scope and introduces frequency domain information and analysis into the overall solution.
[0048] The Fourier transform formula is as follows:
[0049]
[0050] In the formula, F(ω) is the image function of f(t), and f(t) is the image function of F(ω). According to the theory of discrete Fourier transform, using fast Fourier transform to perform Fourier transform of graphic image will be a very efficient and simple mathematical method.
[0051] In general, the two-dimensional FFT change formula for image processing is as follows:
[0052] f(x, y), 0≤x≤M-1, 0≤y≤N-1
[0053] Its Fourier transform kernel is:
[0054]
[0055] The discrete Fourier transform pair of f(x, y) is:
[0056]
[0057]
[0058] ...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a method for distinguishing real and non-real plane images by using frequency spectrum analysis, comprising the following steps: S1, acquiring a two-dimensional image, and inputting the two-dimensional image; S2, performing fast Fourier transform on the two-dimensional image, extracting The amplitude spectrum, phase spectrum, and power spectrum of the image; S3. Analyze the amplitude spectrum. If the energy distribution in the amplitude spectrum is relatively dispersed, it is judged as a non-physical image; if the amplitude spectrum is concentrated in the center, it is judged as a physical image. Compared with other prior art, the present invention has more mature technology, simple algorithm realization and easy integration. The present invention itself still has a lot of sustainability and expansion, does not need to depend on other technologies, and has no technical constraints. The present invention can reuse existing information without additional collection of other information, and the pressure of information processing, transmission and storage is lower. When the invention is applied in fields such as biometric identification, it can realize a high-performance passive detection function without active cooperation of any measured object.
Description
technical field [0001] The invention relates to the field of image recognition, in particular to a method for distinguishing real and non-real plane images by using frequency spectrum analysis. Background technique [0002] There is a common problem in the fields of artificial intelligence, computer vision, biometrics, etc., that is, the replacement of three-dimensional images with non-physical plans or photos often occurs intentionally or unintentionally, and the computer system cannot correctly identify and recognize the problem; the main problem is how to judge Whether the input two-dimensional image is the real object taken or taken again from the photo is very important for some scenes that need to recognize the real object, such as living body recognition. [0003] At present, the recognition of physical objects and non-physical objects is mainly through the use of a large number of depth grayscale information, structured light (information in grayscale two-dimensional...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06T7/00
CPCG06T7/0002G06T2207/30201G06T2207/10004G06T2207/20056
Inventor 罗钧
Owner 上海泛洲信息科技有限公司