CN113658154A - Image detection method and device based on frequency domain periodic texture removal - Google Patents
Image detection method and device based on frequency domain periodic texture removal Download PDFInfo
- Publication number
- CN113658154A CN113658154A CN202110973130.3A CN202110973130A CN113658154A CN 113658154 A CN113658154 A CN 113658154A CN 202110973130 A CN202110973130 A CN 202110973130A CN 113658154 A CN113658154 A CN 113658154A
- Authority
- CN
- China
- Prior art keywords
- image
- periodic texture
- domain
- spectrogram
- frequency domain
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000000737 periodic effect Effects 0.000 title claims abstract description 71
- 238000001514 detection method Methods 0.000 title claims abstract description 59
- 238000001914 filtration Methods 0.000 claims abstract description 45
- 238000004458 analytical method Methods 0.000 claims abstract description 30
- 238000000034 method Methods 0.000 claims abstract description 28
- 238000012545 processing Methods 0.000 claims description 33
- 238000012800 visualization Methods 0.000 claims description 29
- 238000010586 diagram Methods 0.000 claims description 18
- 238000009499 grossing Methods 0.000 claims description 12
- 238000001228 spectrum Methods 0.000 claims description 11
- 238000012216 screening Methods 0.000 claims description 7
- 230000009466 transformation Effects 0.000 claims description 6
- 230000007547 defect Effects 0.000 description 14
- 238000005516 engineering process Methods 0.000 description 7
- 230000000007 visual effect Effects 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012805 post-processing Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000002950 deficient Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000010191 image analysis Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000003711 image thresholding Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000011179 visual inspection Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/168—Segmentation; Edge detection involving transform domain methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10056—Microscopic image
- G06T2207/10061—Microscopic image from scanning electron microscope
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The application provides an image detection method and device based on frequency domain periodic texture removal, wherein the image detection method based on the frequency domain periodic texture removal comprises the following steps: converting the image to be detected from a space domain to a frequency domain through Fourier forward transform, and then converting the image to a power map; obtaining a frequency map after smooth filtering by performing smooth filtering operation on the power map; positioning a signal area corresponding to the periodic texture by using a threshold analysis method; removing a signal area corresponding to the periodic texture to obtain a spectrogram after the periodic texture is removed; and converting the obtained frequency spectrogram after the periodic texture is removed back to a space domain through inverse Fourier transform, and analyzing the output image to obtain a detection result. The method effectively removes periodic textures in the image, clarifies the image, is convenient to identify, and has high detection accuracy and strong universality.
Description
Technical Field
The present application relates to the field of image detection technologies, and in particular, to an image detection method and apparatus based on frequency domain periodic texture removal.
Background
The appearance inspection can find foreign matters, stains, flaws, defects and the like, and prevent the defective products from flowing out, but the visual inspection has a precision limit. The total number detection does not consume labor and cost, and precision deviation and human errors are caused by personal difference. Moreover, fine flaws, stains, etc. are difficult to detect, and therefore, in order to maintain the quality, it is necessary to perform an amplification test by means of a microscope, etc. When the number of dots is small, the microscope can be offline for detection, but when thousands of dots are detected, enormous labor is required, and the production efficiency is greatly reduced. The vision system technology is an indispensable important link to consider quality and production efficiency.
Texture images generally refer to image texture, which is a visual feature reflecting homogeneity in an image and which embodies the tissue arrangement properties of a slowly or periodically changing surface structure of an object surface. Texture has three major landmarks: a local sequence of continuously repeating, non-random arrays, a substantially uniform continuum within a textured area. Texture is different from image features such as gray scale, color, etc., and is represented by the gray scale distribution of pixels and their surrounding spatial neighborhood.
In the field of visual detection, the periodic texture of the image background can make the image blurred, and interference is caused to post-processing flows such as subsequent detection, so that the situations of over-detection and missing detection exist. At present, filtering is performed on an image by using methods such as mean filtering, gaussian filtering and the like, so that the interference of textures can be reduced, but the original defects become fuzzy and are not beneficial to subsequent detection.
Disclosure of Invention
In order to solve the problem that in the prior art, the detection effect is influenced by the interference of periodic textures of an object, the traditional algorithm has high difficulty in design, low detection accuracy and poor universality; the requirement on hardware is high by using a deep learning technology. The method of the invention removes the periodic texture generated in the detected image, thereby clarifying the image, eliminating the interference, and avoiding the problem that the traditional direct filtering makes the whole image fuzzified and is inconvenient for detecting the flaws in the subsequent image.
In a first aspect of the present invention, an image detection method based on frequency domain periodic texture removal is disclosed, which comprises the following steps:
s10, converting the image to be detected from a spatial domain to a frequency domain through two-dimensional discrete Fourier transform, and performing visualization processing to obtain a spectrogram;
s20, converting the frequency spectrum image into a power map for visualization;
s30, performing smooth filtering operation and visualization processing on the power map to obtain a frequency map after smooth filtering;
s40, positioning a signal area corresponding to the periodic texture according to the frequency map after smooth filtering by using a threshold analysis method;
s50, removing a signal area corresponding to the periodic texture to obtain a spectrogram after the periodic texture is removed;
s60, converting the obtained spectrogram after the periodic texture removal back to a spatial domain through two-dimensional inverse discrete Fourier transform, and performing visualization processing to obtain an output graph;
and S70, analyzing the output graph to obtain a detection result.
Preferably, the filtering method in S20 is one of gaussian filtering, binomial filtering and mean filtering.
Preferably, the S40 is:
and performing threshold analysis on the frequency map subjected to smooth filtering, screening out connected domains within a preset area range, and obtaining signal positions corresponding to the periodic textures.
Preferably, the S50 is:
obtaining the position of the maximum power point in the connected domain, and rounding on the frequency graph after smooth filtering by taking the position as the center of a circle and taking a preset value as the radius;
and removing the signal area in the circle to obtain a spectrogram after periodic texture removal.
Preferably, analyzing the output map in S70 includes: and carrying out binarization analysis on the output graph.
Preferably, analyzing the output map in S70 includes: and performing connected domain analysis on the output graph.
Preferably, analyzing the output map in S70 includes: and carrying out linear detection analysis on the output graph.
Preferably, the conversion from the spatial domain to the frequency domain by the fourier transform in S10 is:
the method which combines the base-2, the base-3, the base-4, the base-5, the base-7, the base-8, the base-10, the base-11, the base-13 and the base-16 accelerates the Fourier transform by converting the space domain into the frequency domain through two-dimensional discrete Fourier transform.
Preferably, the visualization processing is to convert the gray scale image into a gray scale image for displaying by normalizing to a gray scale value between 0 and 255.
On the other hand, the present application further provides an image detection apparatus based on frequency domain periodic texture removal, including:
the acquisition spectrogram module converts the image to be detected from a spatial domain to a frequency domain through Fourier forward transform, and performs visualization processing to obtain a spectrogram;
the power map acquisition module is used for converting the frequency spectrum image into a power map and carrying out visualization processing;
the smoothing filtering module is used for performing smoothing filtering operation and visualization processing on the power map to obtain the power map after smoothing filtering;
the positioning texture signal module is used for positioning a signal area corresponding to the periodic texture by using a threshold analysis method according to the smooth filtered power diagram;
the texture removing module is used for removing a signal area corresponding to the periodic texture on the spectrogram to obtain the spectrogram after the periodic texture is removed;
the output image acquisition module converts the obtained frequency domain processing image into a space domain through two-dimensional inverse discrete Fourier transform, and performs visualization processing to obtain an output image;
and the detection analysis module is used for analyzing the output graph to obtain a detection result.
By the scheme, the interference of the periodic texture on the surface of the detected object can be avoided, and the detection of the image is facilitated. According to the scheme disclosed by the embodiment of the invention, the image containing the periodic texture is firstly converted from the spatial domain to the frequency domain to obtain the spectrogram, the composition of signals of the spectrogram is analyzed, the positions of the texture signals corresponding to the spectrogram are compared, the signal part corresponding to the texture is removed, then the frequency spectrum image with the periodic texture removed is converted from the frequency domain to the spatial domain, the image with the texture removed can be obtained, and the output image is analyzed to obtain the detection result.
Drawings
In order to more clearly explain the technical solution of the present application, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious to those skilled in the art that other drawings can be obtained according to the drawings without any creative effort.
FIG. 1 is a flowchart of an image detection method based on frequency domain periodic texture removal according to an embodiment of the present invention;
FIG. 2 is a defect detection input image of a display screen of a mobile phone according to an embodiment of the present invention;
fig. 3 is a frequency spectrum diagram after fourier transform in a defect detection of a mobile phone display screen according to an embodiment of the present invention;
FIG. 4 is a diagram of a frequency domain filter for defect detection of a display screen of a mobile phone according to an embodiment of the present invention;
FIG. 5 is a frequency domain processing diagram after filtering in defect detection of a mobile phone display screen according to an embodiment of the present invention;
FIG. 6 is a Blob identifying defect map of a defect detection post-process of a mobile phone display screen according to an embodiment of the present invention;
fig. 7 is a block diagram of an image detection apparatus based on frequency domain periodic texture removal according to an embodiment of the present invention.
Detailed Description
In order to solve the problem that in the prior art, the detection effect is influenced by the interference of periodic textures of an object, the traditional algorithm has high difficulty in design, low detection accuracy and poor universality; the requirement on hardware is high by using a deep learning technology. The method of the invention removes the periodic texture generated in the detected image, thereby clarifying the image and facilitating the image detection of the flaws therein. The invention discloses an image detection method and device based on frequency domain periodic texture removal through the following embodiments.
Referring to fig. 1, fig. 1 is a flowchart of an image detection method based on frequency domain periodic texture removal, which is applied to a corresponding apparatus. As shown in fig. 1, the method comprises the following steps:
and S10, converting the image to be detected from the spatial domain to the frequency domain through two-dimensional discrete Fourier transform, and performing visualization processing to obtain a spectrogram.
The high and low frequencies of an image are a measure of the intensity variation between locations in the image. Low frequency components, a comprehensive measure of mainly the intensity of the whole image; the high frequency components are mainly a measure of the edges and contours of the image. If the intensities at the positions of one image are equal, the image only has low-frequency components, and only has one main peak and is positioned at the position with zero frequency on the spectrogram of the image. If the intensity of each position of one image changes violently, the image has not only low-frequency components but also a plurality of high-frequency components, and from the spectrum of the image, not only one main peak but also a plurality of side peaks exist. The above phenomenon can be derived by a formula analysis of fourier transform. The two-dimensional image may be decomposed into different frequency components. Where the low frequency component describes a large range of information and the high frequency component describes specific details.
Specifically, the low frequency represents a place where the gradation change is gentle in the image, that is, the background; the high frequencies represent where the grey scale changes are sharp in the image, i.e. the edges. After conversion to the frequency domain, it can be visually seen from the spectrogram that the image has more high frequencies or more low frequencies, the coordinates of the spectrogram correspond to the frequencies, and the brighter places represent the more components of the frequency of the point in the image.
At S10, the spatial domain is transformed into the frequency domain by a two-dimensional discrete fourier transform,
the method which combines the base-2, the base-3, the base-4, the base-5, the base-7, the base-8, the base-10, the base-11, the base-13 and the base-16 accelerates the Fourier transform by converting the space domain into the frequency domain through two-dimensional discrete Fourier transform.
The essence of the fast discrete fourier transform is the discrete fourier transform computation that partitions a long sequence of discrete fourier transform computations into shorter sequences. For the radix-2 algorithm, the sequence is divided into two parts each time, and finally divided into two points DFT, or another division method can be adopted, and the algorithm of radix 3, radix 4, radix 5, etc. is obtained by dividing the sequence into three, four, five, etc. each time.
And S20, converting the spectrogram into a power map and carrying out visualization processing.
The power map is defined as the power map expressed by the conversion of signal power in a unit frequency band in the case of a finite signal and the power changes depending on the frequency, and is the energy expressed by analyzing the available finite signal of the power energy. The spectrogram mainly represents the average transformation of a signal and requires only a period of time for averaging.
The power map is an abbreviation of power spectral density function, which is defined as the signal power within a unit frequency band. It shows the variation of signal power with frequency, i.e. the distribution of signal power in frequency domain. The power map shows the variation of signal power with frequency.
The frequency spectrum diagram and the power diagram are representations of signals, and the frequency spectrum diagram and the power diagram can be converted into signals. It should be noted that the purpose of obtaining the power map in the present application is to find the maximum power point on the power map.
And S30, performing smooth filtering operation and visualization processing on the power map to obtain the power map after smooth filtering.
By analyzing the image, generally the transformation of the defect part on the image is violent and abrupt, then the information of some frequencies is filtered or kept in the frequency domain to enhance the appearance of the defect on the image, and the post-processing can be better carried out to detect the defect. The noise corresponds to the region with higher frequency and the image entity is located in the region with lower frequency.
Therefore, a smoothing filtering operation is required to pass the low frequency component smoothly and effectively block the high frequency component, i.e., to filter the noise of the image and then perform inverse transformation to obtain a smooth image. The smoothing filtering operation is to use one of gaussian filtering, binomial filtering and mean filtering, and then select the most suitable one according to the practical situation.
And S40, according to the smooth filtered power diagram, locating the signal area of the periodic texture by using a threshold analysis method.
And performing threshold analysis on the smooth filtered power diagram, screening out connected domains within a preset area range, and obtaining the signal position of the periodic texture. The core of this step is to smooth the filtered power map to find the signal regions of the periodic texture. Since the signal of the periodic texture is relatively strong reflected in the smoothed power map, the signal region of the periodic texture is localized by threshold segmentation.
The purpose of image thresholding is to divide the set of pixels by gray level, each resulting subset forming a region corresponding to the real scene, each region having consistent properties within it, and adjacent regions having such consistent properties in their layout. Such a division can be achieved by choosing one or more threshold values from the grey scale. And screening connected domains higher than the segmentation threshold value by utilizing image threshold value analysis, and screening through the maximum area and the minimum area, namely, if the area of the connected domain is between the maximum area and the minimum area, keeping the connected domain. It should be noted that the parameters of the threshold analysis need to be determined according to specific practical situations.
And S50, removing the signal region corresponding to the periodic texture on the spectrogram to obtain the spectrogram after the periodic texture is removed.
Obtaining the position of the maximum power point in a connected domain of the smoothly filtered power diagram, and obtaining a signal area in a circle by taking the corresponding position on the frequency spectrum diagram as the center of the circle and a preset value as the radius; and removing the signal area in the circle to obtain a spectrogram after periodic texture removal.
Since the central position of the periodic texture corresponds to the position to the maximum power point, and the bright spots around the maximum power point all represent periodic texture signals, the purpose of S20, S30, and S40 is to find the central position of the periodic texture through the power map, so as to locate the signal regions where the surrounding regions corresponding to the spectrogram are all periodic textures. It should be noted that the area of the periodic texture is analyzed according to the specific actual bright spot density, so as to determine the length of the radius. Step (ii) of
And S60, converting the obtained spectrogram after the periodic texture removal back to a spatial domain through two-dimensional inverse discrete Fourier transform, and performing visualization processing to obtain an output graph.
After the periodic texture on the spectrogram is removed, the filtered image can be obtained after the image is converted back to the spatial domain, and interference generated by the periodic texture in the original image is removed, so that subsequent detection can be better performed.
And S70, analyzing the output graph to obtain a detection result.
And performing post-processing steps such as binarization analysis, connected domain analysis, linear detection and the like on the visualized output image to judge whether flaws exist. Different picture analysis modes are selected according to the actual picture condition, so that the defect part is highlighted, and the detection is more convenient.
In the visualization process, the visualization is a theory, a method and a technology that data is converted into a graph or an image and displayed on a screen by using computer graphics and an image processing technology, and interactive processing is performed. The method specifically comprises the step of converting a spectrogram, a power map, a smooth filtered power map, a frequency domain processing image and an output result image which are obtained in multiple steps into a gray image for display through normalization to 0-255 gray. The output image supports two visual display modes of real and byte. This way the presented image content of each step can be displayed most intuitively.
In a specific example, a detailed application and solution of the method is provided. In the development process of science and technology, the quality requirement of people on a mobile phone screen is higher and higher, generally speaking, the more colors that can be displayed on the mobile phone screen, the more complex images can be displayed, and the more rich the hierarchy of the displayed images. Therefore, in order to satisfy better use effect, manufacturers need to correspondingly detect the performance of the display screen of the mobile phone and the functions of the mobile phone before the mobile phone enters the market.
Referring to fig. 2, an image of a display screen of a mobile phone with a defective portion is shown. Because the color difference is not large, the periodic ripples occupy a large proportion in the whole image and are difficult to distinguish, some image detection analysis is needed by converting the image, and then flaws are obviously identified.
Firstly, reading an original mobile phone screen partial image, wherein the size of the image is M x N, M represents the line number of the image and N represents the column number of the image, and then performing two-dimensional discrete Fourier transform on the image by using a formula (1). The transformation process adopts the method of fusing base-2, base-3, base-4, base-5, base-7, base-8, base-10, base-11, base-13, base-16 and the like to accelerate Fourier transformation.
Where x is 0,1,2, …, M-1, y is 0,1,2, …, N-1, u, v denote frequency domain coordinates, and j denotes an imaginary unit.
Then, the amplitude of the frequency is obtained by calculating the modulus of F (u, v), as shown in formula (2):
wherein R represents the real part of F (u, v) and I represents the imaginary part of F (u, v).
Then, after normalization through visualization operation, a spectrogram of the mobile phone flaw detection is finally obtained, as shown in fig. 3.
Then, converting the frequency spectrogram of the mobile phone flaw detection into a power chart of the mobile phone flaw detection, further performing smooth filtering operation on the power chart | F (u, v) | of the mobile phone flaw detection by using a Gaussian filtering, binomial filtering or mean filtering method to obtain a power chart | F '(u, v) | after smooth filtering, then setting a segmentation threshold value T, and screening a point P (u, v) meeting the condition | F' (u, v) | > T; analyzing the points to obtain several connected domains S, calculating the area of each connected domain, and setting the maximum area SmaxMinimum area SminScreening out satisfaction Smin＜S＜SmaxThe region S'. After visualizing | F '(u, v) | and the region S', the result is as shown in fig. 4.
Calculating the point with the maximum median value of each block of region S' in the power map | F (u, v) | after smooth filtering, and marking as Pmax(u, v), taking the point as a center of circle, setting a radius as R to make a circle C, and setting the real part and the imaginary part of each point in the corresponding spectrogram F (u, v) in the circle as 0, namely F (u, v) is 0, u, v belongs to C. When the circle inner point exceeds the image portion, no processing is performed, and after the processed result F (u, v) is visualized, the result is shown in fig. 5.
And (3) performing two-dimensional inverse discrete Fourier transform on the frequency domain processed image F (u, v) by using a formula (3), wherein the Fourier transform is accelerated by adopting methods of fusing a base-2, a base-3, a base-4, a base-5, a base-7, a base-8, a base-10, a base-11, a base-13, a base-16 and the like.
Where x is 0,1,2, …, M-1, y is 0,1,2, …, N-1, u, v denote frequency domain coordinates, and j denotes an imaginary unit.
Then, taking the real part of f' (x, y), as shown in equation (4):
|f′(x,y)|＝r(x,y)(4)
where r (x, y) represents the real part of f' (x, y).
And normalizing through visual operation to finally obtain an output image.
The resulting output map was analyzed for connected components using the Blob tool to obtain a defect-labeled map 6, in which the middle white region indicates a defect.
It should be noted that Blob analysis aims to detect and analyze 2-D shapes in an image, and to obtain information such as the position, shape, orientation, and topological relationship between objects, i.e., inclusion relationship. From this information, the target can be identified. In some applications we need to use not only the shape features of 2D, but also the feature relationships between Blob analyses.
The Blob analysis employed in the present embodiment is not limited to this analysis tool, as the image analysis. Referring to fig. 7, the flaws can be identified by the eyes of ordinary people, and the essence of the flaws is that the image is differentiated and clarified by the method of the present application, so that the characteristics of the image can be amplified more obviously, and then the flaws can be automatically identified by an image analysis algorithm.
Further, referring to fig. 7, an image detection apparatus based on frequency domain periodic texture removal disclosed in the embodiment of the present invention further includes:
the acquisition spectrogram module converts the image to be detected from a spatial domain to a frequency domain through Fourier forward transform, and performs visualization processing to obtain a spectrogram;
the power map acquisition module is used for converting the frequency spectrum image into a power map and carrying out visualization processing;
the smoothing filtering module is used for performing smoothing filtering operation and visualization processing on the power map to obtain the power map after smoothing filtering;
the positioning texture signal module is used for positioning a signal area corresponding to the periodic texture by using a threshold analysis method according to the smooth filtered power diagram;
the texture removing module is used for removing a signal area corresponding to the periodic texture on the spectrogram to obtain the spectrogram after the periodic texture is removed;
the output image acquisition module converts the obtained frequency domain processing image into a space domain through two-dimensional inverse discrete Fourier transform, and performs visualization processing to obtain an output image;
and the detection analysis module is used for analyzing the output graph to obtain a detection result.
In the previous description, numerous specific details were set forth in order to provide a thorough understanding of the present invention. The foregoing description is only a preferred embodiment of the invention, which can be embodied in many different forms than described herein, and therefore the invention is not limited to the specific embodiments disclosed above. And that those skilled in the art may, using the methods and techniques disclosed above, make numerous possible variations and modifications to the disclosed embodiments, or modify equivalents thereof, without departing from the scope of the claimed embodiments. Any simple modification, equivalent change and modification of the above embodiments according to the technical essence of the present invention are within the scope of the technical solution of the present invention.
Claims (10)
1. An image detection method based on frequency domain periodic texture removal is characterized by comprising the following steps:
s10, converting the image to be detected from a spatial domain to a frequency domain through two-dimensional discrete Fourier transform, and performing visualization processing to obtain a spectrogram;
s20, converting the spectrogram into a power map and carrying out visualization processing;
s30, performing smooth filtering operation and visualization processing on the power map to obtain the power map after smooth filtering;
s40, positioning a signal area of the periodic texture by using a threshold analysis method according to the smooth filtered power map;
s50, removing a signal area corresponding to the periodic texture on the spectrogram to obtain the spectrogram after the periodic texture is removed;
s60, converting the obtained spectrogram after the periodic texture removal back to a spatial domain through two-dimensional inverse discrete Fourier transform, and performing visualization processing to obtain an output graph;
and S70, analyzing the output graph to obtain a detection result.
2. The method of claim 1, wherein the smoothing filtering operation in S30 is a filtering method using one of gaussian filtering, binomial filtering and mean filtering.
3. The method for detecting an image based on frequency-domain periodic texture removal according to claim 2, wherein said S40 is:
and performing threshold analysis on the smooth filtered power diagram, screening out connected domains within a preset area range, and obtaining the signal position of the periodic texture.
4. The method according to claim 3, wherein said S50 is:
obtaining the position of the maximum power point in a connected domain of the smoothly filtered power diagram, and obtaining a signal area in a circle by taking the corresponding position on the frequency spectrum diagram as the center of the circle and a preset value as the radius;
and removing the signal area in the circle to obtain a spectrogram after periodic texture removal.
5. The method for detecting the image based on the frequency-domain periodic texture removal according to any one of claims 1 to 4, wherein the analyzing the output map in the S70 is to perform binarization analysis on the output map.
6. The image detection method based on frequency domain periodic texture removal according to any one of claims 1 to 4, wherein the analyzing of the output map in S70 is to perform connected domain analysis on the output map.
7. The image detection method based on frequency domain periodic texture removal according to any one of claims 1 to 4, wherein the analyzing of the output map in S70 is to perform straight line detection analysis on the output map.
8. The image detection method based on frequency-domain periodic texture removal according to any one of claims 1 to 7, wherein the transformation from the spatial domain to the frequency domain by the forward fourier transform in S10 is:
the method which combines the base-2, the base-3, the base-4, the base-5, the base-7, the base-8, the base-10, the base-11, the base-13 and the base-16 accelerates the Fourier transform by converting the space domain into the frequency domain through two-dimensional discrete Fourier transform.
9. The image detection method based on frequency domain periodic texture removal according to any one of claims 1 to 8,
the visualization processing is to convert the gray scale image into a gray scale image for display by normalizing the gray scale value to be between 0 and 255.
10. An image detection device based on frequency domain periodic texture removal, comprising:
the acquisition spectrogram module converts the image to be detected from a spatial domain to a frequency domain through Fourier forward transform, and performs visualization processing to obtain a spectrogram;
the power map acquisition module is used for converting the frequency spectrum image into a power map and carrying out visualization processing;
the smoothing filtering module is used for performing smoothing filtering operation and visualization processing on the power map to obtain the power map after smoothing filtering;
the positioning texture signal module is used for positioning a signal area corresponding to the periodic texture by using a threshold analysis method according to the smooth filtered power diagram;
the texture removing module is used for removing a signal area corresponding to the periodic texture on the spectrogram to obtain the spectrogram after the periodic texture is removed;
the output image acquisition module converts the obtained frequency domain processing image into a space domain through two-dimensional inverse discrete Fourier transform, and performs visualization processing to obtain an output image;
and the detection analysis module is used for analyzing the output graph to obtain a detection result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110973130.3A CN113658154A (en) | 2021-08-24 | 2021-08-24 | Image detection method and device based on frequency domain periodic texture removal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110973130.3A CN113658154A (en) | 2021-08-24 | 2021-08-24 | Image detection method and device based on frequency domain periodic texture removal |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113658154A true CN113658154A (en) | 2021-11-16 |
Family
ID=78492622
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110973130.3A Pending CN113658154A (en) | 2021-08-24 | 2021-08-24 | Image detection method and device based on frequency domain periodic texture removal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113658154A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114998112A (en) * | 2022-04-22 | 2022-09-02 | 南通悦福软件有限公司 | Image denoising method and system based on adaptive frequency domain filtering |
| CN115294015A (en) * | 2022-06-20 | 2022-11-04 | 中国石油大学(北京) | Method and equipment for determining shale striated layer and storage medium |
| CN116548928A (en) * | 2023-07-11 | 2023-08-08 | 西安浩阳志德医疗科技有限公司 | Nursing service system based on internet |
| CN117314798A (en) * | 2023-11-29 | 2023-12-29 | 国药新光医疗科技有限公司 | Method for removing fiber mirror image grid |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005115548A (en) * | 2003-10-06 | 2005-04-28 | Nippon Telegr & Teleph Corp <Ntt> | Image processor, image processing method, image processing program, and recording medium recorded with it program |
| US20130272616A1 (en) * | 2011-12-29 | 2013-10-17 | Marat Ravilevich Gilmutdinov | Method of and apparatus for low-complexity detection of periodic textures orientation |
| CN108961225A (en) * | 2018-06-21 | 2018-12-07 | 重庆信息通信研究院 | Periodic texture suppressing method and device |
| CN109507198A (en) * | 2018-12-14 | 2019-03-22 | 航天科工智能机器人有限责任公司 | Mask detection system and method based on Fast Fourier Transform (FFT) and linear Gauss |
| CN110084768A (en) * | 2019-05-10 | 2019-08-02 | 江南大学 | The defect inspection method of LCD light guide plate based on background filtering |
| CN111080636A (en) * | 2019-12-23 | 2020-04-28 | 杨静日 | CNN semantic segmentation self-learning detection method for surface defects of color steel tiles |
| CN111383186A (en) * | 2018-12-29 | 2020-07-07 | Tcl集团股份有限公司 | Image processing method and device and terminal equipment |
-
2021
- 2021-08-24 CN CN202110973130.3A patent/CN113658154A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005115548A (en) * | 2003-10-06 | 2005-04-28 | Nippon Telegr & Teleph Corp <Ntt> | Image processor, image processing method, image processing program, and recording medium recorded with it program |
| US20130272616A1 (en) * | 2011-12-29 | 2013-10-17 | Marat Ravilevich Gilmutdinov | Method of and apparatus for low-complexity detection of periodic textures orientation |
| CN108961225A (en) * | 2018-06-21 | 2018-12-07 | 重庆信息通信研究院 | Periodic texture suppressing method and device |
| CN109507198A (en) * | 2018-12-14 | 2019-03-22 | 航天科工智能机器人有限责任公司 | Mask detection system and method based on Fast Fourier Transform (FFT) and linear Gauss |
| CN111383186A (en) * | 2018-12-29 | 2020-07-07 | Tcl集团股份有限公司 | Image processing method and device and terminal equipment |
| CN110084768A (en) * | 2019-05-10 | 2019-08-02 | 江南大学 | The defect inspection method of LCD light guide plate based on background filtering |
| CN111080636A (en) * | 2019-12-23 | 2020-04-28 | 杨静日 | CNN semantic segmentation self-learning detection method for surface defects of color steel tiles |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114998112A (en) * | 2022-04-22 | 2022-09-02 | 南通悦福软件有限公司 | Image denoising method and system based on adaptive frequency domain filtering |
| CN115294015A (en) * | 2022-06-20 | 2022-11-04 | 中国石油大学(北京) | Method and equipment for determining shale striated layer and storage medium |
| CN115294015B (en) * | 2022-06-20 | 2023-12-01 | 中国石油大学(北京) | Shale layer determination method, shale layer determination equipment and shale layer storage medium |
| CN116548928A (en) * | 2023-07-11 | 2023-08-08 | 西安浩阳志德医疗科技有限公司 | Nursing service system based on internet |
| CN116548928B (en) * | 2023-07-11 | 2023-09-08 | 西安浩阳志德医疗科技有限公司 | Nursing service system based on internet |
| CN117314798A (en) * | 2023-11-29 | 2023-12-29 | 国药新光医疗科技有限公司 | Method for removing fiber mirror image grid |
| CN117314798B (en) * | 2023-11-29 | 2024-05-03 | 国药新光医疗科技有限公司 | Method for removing fiber mirror image grid |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113658154A (en) | Image detection method and device based on frequency domain periodic texture removal | |
| Yu et al. | A machine vision method for measurement of machining tool wear | |
| CN113643271A (en) | Image flaw detection method and device based on frequency domain filtering | |
| Wang et al. | Chromosome image enhancement using multiscale differential operators | |
| Liu et al. | Fabric defect detection based on information entropy and frequency domain saliency | |
| CN111161222B (en) | Printing roller defect detection method based on visual saliency | |
| Flores et al. | Breast ultrasound despeckling using anisotropic diffusion guided by texture descriptors | |
| CN109801286B (en) | Surface defect detection method for LCD light guide plate | |
| CN116137036B (en) | Gene detection data intelligent processing system based on machine learning | |
| CN112614062A (en) | Bacterial colony counting method and device and computer storage medium | |
| CN116468726B (en) | Online foreign matter line detection method and system | |
| Peng et al. | Automated product boundary defect detection based on image moment feature anomaly | |
| Fonseca-Pinto et al. | A textured scale-based approach to melanocytic skin lesions in dermoscopy | |
| Huang | Improved non-local means algorithm for image denoising | |
| Estrozi et al. | 1D and 2D Fourier-based approaches to numeric curvature estimation and their comparative performance assessment | |
| An et al. | Image processing algorithm based on bi-dimensional local mean decomposition | |
| Zhu et al. | Quantitative assessment mechanism transcending visual perceptual evaluation for image dehazing | |
| Shao et al. | No-Reference image quality assessment based on edge pattern feature in the spatial domain | |
| Aqeel | The Use of Threshold Technique in image segmentation | |
| Xiao et al. | 3D center segregation reconstruction of steel continuous casting slab | |
| Zhang et al. | Using multiscale structural fusion for sonar image quality evaluation | |
| Zhang et al. | A fast learning-based super-resolution method for copper strip defect image | |
| CN110766672B (en) | Electrode fouling detection method | |
| US11025924B2 (en) | Image processing method and recording medium | |
| Lin et al. | A noise level estimation method of impulse noise image based on local similarity |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |