CN110610525B - Image processing method and device and computer readable storage medium - Google Patents

Image processing method and device and computer readable storage medium Download PDF

Info

Publication number
CN110610525B
CN110610525B CN201810621822.XA CN201810621822A CN110610525B CN 110610525 B CN110610525 B CN 110610525B CN 201810621822 A CN201810621822 A CN 201810621822A CN 110610525 B CN110610525 B CN 110610525B
Authority
CN
China
Prior art keywords
image
characteristic data
illumination
processed
determining
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.)
Active
Application number
CN201810621822.XA
Other languages
Chinese (zh)
Other versions
CN110610525A (en
Inventor
李德志
王开
赵玺
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
ZTE Corp
Original Assignee
ZTE Corp
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by ZTE Corp filed Critical ZTE Corp
Priority to CN201810621822.XA priority Critical patent/CN110610525B/en
Publication of CN110610525A publication Critical patent/CN110610525A/en
Application granted granted Critical
Publication of CN110610525B publication Critical patent/CN110610525B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/90Determination of colour characteristics
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10024Color image
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20024Filtering details
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20048Transform domain processing
    • G06T2207/20056Discrete and fast Fourier transform, [DFT, FFT]

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Image Processing (AREA)

Abstract

The invention discloses an image processing method, which comprises the following steps: carrying out color space processing on an image to be processed to obtain brightness characteristic data of the image to be processed; carrying out low-pass filtering processing on the brightness characteristic data to obtain illumination characteristic data of the image to be processed; determining an adjusting parameter through a set reference gray value and the illumination characteristic data; and performing illumination removing processing on the brightness characteristic data according to the adjustment parameters and the illumination characteristic data to obtain a target image. The invention also discloses an image processing device and a computer readable storage medium.

Description

Image processing method and device and computer readable storage medium
Technical Field
The embodiment of the invention relates to the field of communication, in particular to but not limited to a method and a device for image processing and a computer-readable storage medium.
Background
With the increase of network scale, the related technology of image recognition, especially the face recognition technology, has made great progress at present, but in an uncontrollable environment, the image can be affected by factors such as illumination, expression, individual difference and the like. When the factors are not obviously changed in different images, the recognition rate of the images reaches over 90 percent; however, when these factors of the image are changed, the recognition rate of the image is significantly reduced, so that the application range of the image recognition technology is also limited.
Among the above factors, the variation of illumination includes the variation of intensity and the variation of angle, and the variation of intensity causes the occurrence of extreme illumination conditions such as dim light, high light and the like; the change in angle causes the image to appear with different degrees of shadow regions, which degrades the recognition rate of the image, especially the facial textural features of the human face. Therefore, the influence of illumination on image recognition is not negligible, and especially under the condition of complicated outdoor illumination change, even the recognition rate of the image recognition method with high recognition rate is seriously reduced, and visible illumination change becomes one of the bottlenecks restricting the development of the image recognition related technology.
Disclosure of Invention
In view of the foregoing, it is desirable to provide an image processing method, an image processing apparatus, and a computer-readable storage medium, which are capable of removing the influence of external illumination, performing orderly convergence on an image from which the external illumination is removed, and reducing the influence of illumination conditions on image recognition.
The technical scheme of the embodiment of the invention is realized as follows:
in one aspect, an embodiment of the present invention provides an image processing method, including:
carrying out color space processing on an image to be processed to obtain lightness characteristic data of the image to be processed;
carrying out low-pass filtering processing on the brightness characteristic data to obtain illumination characteristic data of the image to be processed;
determining an adjustment parameter through a set reference gray value and the illumination characteristic data;
and performing illumination removing processing on the brightness characteristic data according to the adjustment parameters and the illumination characteristic data to obtain a target image.
In another aspect, an embodiment of the present invention provides an image processing apparatus, including: the device comprises a color processing unit, a filtering unit, a determining unit and a target unit; wherein, the first and the second end of the pipe are connected with each other,
the color processing unit is used for performing color space processing on an image to be processed to obtain lightness characteristic data of the image to be processed;
the filtering unit is used for carrying out low-pass filtering processing on the brightness characteristic data to obtain illumination characteristic data of the image to be processed;
the determining unit is used for determining an adjusting parameter according to a set reference gray value and the illumination characteristic data;
and the target unit is used for carrying out illumination removing processing on the lightness characteristic data according to the adjusting parameters and the illumination characteristic data to obtain a target image.
In one aspect, an embodiment of the present invention provides an image processing apparatus, including: a processor and a memory for storing a computer program operable on the processor, wherein the processor is adapted to perform the steps of the image processing method described above when running the computer program.
In still another aspect, an embodiment of the present invention provides a computer-readable storage medium, on which an image processing program is stored, and the image processing program, when executed by a processor, implements the steps of the image processing method described above.
The image processing method, the image processing device and the computer-readable storage medium of the embodiment of the invention perform color space processing on an image to be processed to obtain lightness characteristic data of the image to be processed; carrying out low-pass filtering processing on the brightness characteristic data to obtain illumination characteristic data of the image to be processed; determining an adjusting parameter through a set reference gray value and the illumination characteristic data; performing illumination removing processing on the brightness characteristic data according to the adjustment parameters and the illumination characteristic data to obtain a target image; in this way, the external illumination of the image is removed through low-pass filtering processing, the image without the external illumination is converged to a uniform gray value through adjusting parameters so as to perform orderly convergence, all the images have similar brightness, the influence of illumination conditions on image recognition is reduced, and the recognition rate of the image recognition is improved.
Drawings
Fig. 1 is a schematic flowchart of an image processing method according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating an image processing method according to a second embodiment of the present invention;
fig. 3 is a schematic flowchart of a processing method for generating an image according to a fourth embodiment of the present invention;
fig. 4 is a schematic structural diagram of an image processing system according to a fifth embodiment of the present invention;
fig. 5 is a first schematic view illustrating an effect of the image processing method according to the fifth embodiment of the present invention;
fig. 6 is a schematic diagram illustrating an effect of the image processing method according to the fifth embodiment of the present invention;
fig. 7 is a schematic structural diagram of an image processing apparatus according to a sixth embodiment of the present invention;
fig. 8 is a schematic structural diagram of another image processing apparatus according to a sixth embodiment of the present invention;
fig. 9 is a schematic structural diagram of an image processing apparatus according to an eighth embodiment of the present invention.
Detailed Description
The following describes the implementation of the technical solution in further detail with reference to the accompanying drawings.
Example one
An embodiment of the present invention provides an image processing method, as shown in fig. 1, the method includes:
s101, performing color space processing on an image to be processed to obtain brightness characteristic data of the image to be processed;
the image to be processed is a Red-Green-Blue (RGB) image, and the RGB image is subjected to color space processing to obtain a Hue-Saturation-Value (HSV) image corresponding to the image to be processed, wherein the Hue-Saturation-Value (HSV) image is obtained. The parameters of the color of the HSV image are respectively: hue (Hue, H), saturation (S), lightness (Value, V). And after obtaining the HSV image, extracting a parameter V in the HSV image to obtain brightness characteristic data of the image to be processed, wherein the brightness characteristic data of the image to be processed represents the brightness degree of the color of the image to be processed.
S102, carrying out low-pass filtering processing on the lightness characteristic data to obtain illumination characteristic data of the image to be processed;
after the brightness characteristic data of the image to be processed is determined in S101, the brightness characteristic data is processed through a low-pass filter, and the illumination characteristic data of the image to be processed is obtained. The low-pass filter may be a gaussian filter, a butterworth filter, or the like.
The illumination factors of the image comprise internal illumination factors and external illumination factors, the external illumination factors are small in change and mainly distributed in a low-frequency area of the image; the variation range of internal illumination factors such as invariant features in the image is large, and the internal illumination factors are mainly distributed in a high-frequency area of the image. Processing the brightness characteristic data through the low-pass filter, reserving external illumination factors of the image, and removing the internal illumination factors of the image, namely, performing low-pass filtering processing on the brightness characteristic data of the image to be processed to obtain illumination characteristic data which is the image data without the internal illumination factors, namely the external illumination data.
In an embodiment, the low-pass filtering the brightness feature data to obtain the illumination feature data of the image to be processed includes: performing Fourier transform on the lightness characteristic data to obtain a first frequency spectrum distribution corresponding to the lightness characteristic data; filtering the first frequency spectrum distribution through a low-pass filter to obtain a second frequency spectrum distribution; and carrying out Fourier inverse transformation on the second frequency spectrum distribution to obtain the illumination characteristic data of the image to be processed.
Here, the lightness feature data is fourier-transformed to obtain a first spectral distribution in the frequency domain, which is the spectral distribution of the lightness feature data. The first spectral distribution is a spectral distribution of the lightness feature data in the entire frequency domain. And performing low-pass filtering processing on the first frequency spectrum distribution through a low-pass filter to filter the frequency spectrum distribution of the first frequency spectrum distribution in a high frequency band, so as to obtain the frequency spectrum distribution of the first frequency spectrum distribution in a low frequency band, namely a second frequency spectrum distribution. At this time, the internal illumination factor of the image to be processed is removed.
Before the first spectral distribution is filtered by the low-pass filter to obtain a second spectral distribution, the cut-off frequency of the low-pass filter is determined. Here, determining the cutoff frequency of the low pass filter may include: determining an energy weight according to the image to be processed; determining a lower limit frequency point and an upper limit frequency point of the first frequency spectrum distribution; determining first energy of the first spectrum distribution at the lower limit frequency point and the upper limit frequency point; determining second energy of the first frequency spectrum distribution at the lower limit frequency point and the characteristic frequency point; and if the ratio of the first energy to the second energy is an energy weight, determining the characteristic frequency point corresponding to the second energy as the cutoff frequency of the low-pass filter. The first energy is the overall energy of the first spectrum distribution, and the second energy is the energy of the first spectrum distribution from the lower limit frequency point to the characteristic frequency point. When the characteristic frequency point is determined, the value of the characteristic frequency point can be increased gradually from 0 or a lower limit frequency point in an iterative manner until the ratio of the second energy to the first energy is an energy weight, and the characteristic frequency point corresponding to the second energy at the moment is used as the cutoff frequency of the low-pass filter. Here, the lower limit frequency point is the lowest frequency point of the first spectrum distribution, and the upper limit frequency point is the highest frequency point of the second spectrum distribution.
It should be noted that the selection of the cutoff frequency of the low-pass filter is related to the brightness feature data itself of the image to be processed, and therefore, the cutoff frequencies of the low-pass filters that perform the low-pass filtering processing on different images to be processed are independent of each other.
The above process of determining the energy weight used in determining the cut-off frequency of the low-pass filter may include: carrying out histogram equalization processing on the image to be processed to obtain an equalized image; carrying out color space processing on the balanced image to obtain balanced lightness characteristic data of the balanced image; and determining the energy weight according to the brightness characteristic data of the image to be processed and the balanced brightness characteristic data.
Performing HE (Histogram Equalization) processing on an image to be processed by using an HE (height Equalization) adjustment algorithm to obtain an equalized image of the image to be processed, performing color space processing on the equalized image to obtain an HSV (hue, saturation and value) image corresponding to the equalized image, and extracting a parameter V of the HSV image corresponding to the equalized image to obtain equalized brightness characteristic data; and determining the ratio of the energy of the brightness characteristic data of the image to be processed to the energy of the balanced brightness characteristic data as an energy weight.
It should be noted that, for different images, the illumination conditions are different, and the energy of the external illumination, i.e., the second energy, is different, so that for the images under different illumination conditions, the values of the energy weights are different.
S103, determining an adjusting parameter through a set reference gray value and the illumination characteristic data;
the set reference gray value can represent a fixed image without external illumination factors, and the value range of the set reference gray value is 0-255, for example: 40. 60 or 90. And obtaining a target image aligned with the brightness of the reference gray image by approximating the reference gray value and the image of the image to be processed without the external illumination. At this time, the target image can be characterized by a target self-quotient graph, and the target self-quotient graph is determined by adjusting the parameters and the illumination characteristic data as well as the brightness characteristic data of the image to be processed.
In an embodiment, the tuning parameters include a scaling parameter and an offset tuning parameter; correspondingly, the determining of the adjustment parameter by the set reference gray value and the illumination characteristic data includes:
determining a target self-quotient graph T according to formula (1):
Figure GDA0003888757530000061
wherein alpha is a scale adjustment parameter, beta is an offset adjustment parameter,
Figure GDA0003888757530000071
is the illumination characteristic data, and I is the lightness characteristic data; determining a two-norm between the reference gray value and the target self-quotient graph T; and minimizing the two norms, and converging the image to be processed to the image corresponding to the reference gray value to determine the scale adjustment parameter and the offset adjustment parameter. Wherein the two norms are the sum of squares of all elements and the root number.
And S104, performing illumination removing processing on the brightness characteristic data according to the adjusting parameters and the illumination characteristic data to obtain a target image.
After the adjustment parameters are determined in S103, a target self-quotient graph is calculated through the adjustment parameters, the illumination characteristic data obtained through the low-pass filtering process, and the brightness characteristic data, and the calculated target self-quotient graph is determined as a target image. Here, the target self-quotient graph is obtained by dividing the brightness characteristic data by the illumination characteristic data by using the adjustment parameter as a coefficient in the de-illumination processing.
In an embodiment, the performing, according to the adjustment parameter and the illumination characteristic data, a de-illumination process on the brightness characteristic data to obtain a target image includes:
determining a target self-quotient graph T according to formula (1):
Figure GDA0003888757530000072
wherein alpha is a scale adjustment parameter, beta is an offset adjustment parameter,
Figure GDA0003888757530000073
is the illumination characteristic data, and I is the lightness characteristic data; and taking the target self-quotient graph as the target image.
It should be noted that the adjustment parameter in the embodiment of the present invention may be set by an adjustment policy according to the image to be processed and the set reference gray value, and in the above description, the adjustment policy is described by setting the adjustment parameter including the scale adjustment parameter and the offset adjustment parameter, and the adjustment policy is not specifically limited in the embodiment of the present invention.
In this embodiment, color space processing is performed on an image to be processed to obtain lightness feature data of the image to be processed; carrying out low-pass filtering processing on the brightness characteristic data to obtain illumination characteristic data of the image to be processed; determining an adjustment parameter through a set reference gray value and the illumination characteristic data; and performing illumination removing processing on the brightness characteristic data according to the adjustment parameters and the illumination characteristic data to obtain a target image. In this way, the external illumination of the image is removed through low-pass filtering processing, the image without the external illumination is converged to a uniform gray value through adjusting parameters so as to perform orderly convergence, all the images have similar brightness, the influence of illumination conditions on image recognition is reduced, and the recognition rate of the image recognition is improved.
The image processing method provided by the embodiment of the invention has the following advantages when the external illumination removal is carried out on the image:
(1) And converging the whole aligned target image to a predefined reference gray value, unifying the whole illumination condition and enabling all the aligned target images to have similar brightness. The reference gray value is a constant, and the corresponding texture image has symmetry.
(2)、Adjusting the image corresponding to the illumination characteristic data by adopting the adjustment parameters
Figure GDA0003888757530000081
The contrast and compensation in the image processing system enable the illumination condition of the target image to be more symmetrical.
(3) And for the images under different illumination conditions, during the low-pass filtering processing, according to the difference of the frequency spectrum distribution of the image to be processed, the cut-off frequency of the low-pass filter is dynamically determined, so that the illumination characteristic data is more accurate.
Example two
In the embodiment of the present invention, the method for processing an image according to the embodiment of the present invention is further described, as shown in fig. 2, including:
s201, performing color space processing on an image to be processed to obtain brightness characteristic data of the image to be processed;
converting the image to be processed into an HSV (hue, saturation and value) color space through the color space to obtain an HSV image corresponding to the image to be processed, and taking the light and dark components of a brightness V channel of the HSV image to obtain brightness characteristic data I.
S202, carrying out Fourier transform on lightness characteristic data to obtain first frequency spectrum distribution;
here, the lightness feature data image is converted into the frequency domain using two-dimensional fast fourier transform, and the corresponding frequency spectrum distribution, i.e., the first frequency offset distribution F is obtained g
S203, determining the cut-off frequency of the low-pass filter according to the first spectrum distribution and the energy weight;
here, in conjunction with the first spectral distribution of the image to be processed, the cutoff frequency of the low-pass filter used when the low-pass filtering processing is performed on the first spectral distribution is determined in accordance with the preset energy weight P to determine the low-pass filter.
S204, low-pass filtering is carried out on the first frequency spectrum distribution through a low-pass filter to obtain a second frequency spectrum distribution;
here, the second spectral distribution is denoted as F ″ g
S205, performing inverse Fourier transform on the second frequency spectrum distribution to obtain illumination characteristic data of the image to be processed;
here, the second spectral distribution F' after the low-pass filtering is filtered g Performing inverse Fourier transform to obtain illumination characteristic data representing external illumination factors of the image to be processed
Figure GDA0003888757530000091
S206, minimizing the two norms of the gray value of the reference image and the target self-quotient graph, and estimating to obtain an adjusting parameter;
the self-adjusting parameters comprise a scale adjusting parameter and an offset adjusting parameter, the scale adjusting parameter is used for correspondingly scaling and adjusting the illumination characteristic data, and the offset adjusting parameter is used for carrying out offset adjustment on the illumination characteristic data so as to adjust the contrast and compensation change of the external illumination condition corresponding to the illumination characteristic data when determining the target self-quotient graph of the target image.
The adjustment parameters are determined by equation (2):
argmin||I p -T|| F (2)
wherein, I p For reference gray value, target self-quotient graph
Figure GDA0003888757530000092
The adjustment parameters α and β are thus determined by minimizing the two-normal form of the reference gray value and the target self-quotient graph.
And S207, determining a target image according to the adjusting parameters.
And calculating to obtain a target self-quotient graph through the adjustment parameters alpha and beta of the stages, and taking the target self-quotient graph as a target image.
EXAMPLE III
In the embodiment of the present invention, an image to be processed is taken as an example of a face image, and the image processing method provided in the embodiment of the present invention is further described.
Before explaining the image processing method provided by the embodiment of the present invention, the technique related to the self-quotient involved in the embodiment of the present invention is described.
In the related art, the self-quotient of an image is defined based on a Lambertian optical model. The lambertian optical model is an ideal diffuse reflection model, and assuming that the surface of an object has uniform scattering in all directions, the reflection model B is expressed by formula (3):
B(x,y)=ρ(x,y)*S(x,y)*L(x,y) (3)
where S (x, y) represents a normal vector of the object surface, ρ (x, y) represents the reflectance of the object surface at the point (x, y), and L (x, y) represents the illumination condition. Obviously, under the assumption of the lambertian optical model, the intrinsic property emissivity and the surface normal vector of the human face and the external illumination factor are included in the human face imaging process. Defining a self-quotient graph Q (x, y) = ρ (x, y) × S (x, y), equation (3) is converted into equation (4):
B(x,y)=Q(x,y)*L(x,y) (4)
here, the internal illumination invariant characteristic Q (x, y) of the image is directly separated from the external illumination factor L (x, y). In the lambertian model, the external illumination components in the image are small in change and mainly distributed in a low-frequency area of the image, and the internal characteristics Q (x, y) which embody the face identity information are large in change amplitude at the boundary of the face area and correspond to high-frequency components in the image.
On the premise of a lambertian optical model, a self-quotient graph method in the related art is as follows:
the influence of external illumination is removed through a low-pass filtering kernel, invariant features of the internal illumination of the human face (corresponding to Q (x, y) in the formula (4), namely a self-quotient graph) are extracted, and illumination normalization is achieved.
The definition of the self-quotient graph Q is as follows:
Figure GDA0003888757530000101
wherein the content of the first and second substances,
Figure GDA0003888757530000102
is image data (illumination characteristic data) after the image I is subjected to smoothing filtering processing, and F is a filtering kernel of a low-pass filter for reducing the influence of external illumination. Wherein the low-pass filterMay be composed of a plurality of different filter kernels.
The self-quotient graph method in the related technology has certain problems, the self-quotient graph represents the essential attribute of the face image I, the calculation of the self-quotient graph is calculated point by taking pixel points (x, y) as a unit, the selection of a filtering kernel F is very critical, and how to determine a filter is related to the effect of light alignment. In the related technology, a fixed Gaussian filter is adopted as a filtering kernel F, and the illumination problem in a face recognition system can be effectively solved by the method for determining the self-quotient graph.
In the image processing method provided by the embodiment of the invention, the self-quotient graph is improved, a concept of self-luminous ratio is provided, and the target self-quotient graph T of the image is as follows:
Figure GDA0003888757530000111
wherein L is R Denotes the self-illumination ratio,. Epsilon.denotes the low-pass filter,. Tau.and.tau -1 The fourier transform and inverse fourier transform processes are shown separately. T denotes the processed target self-quotient graph,
Figure GDA0003888757530000112
representing a smoothed image of image I.
It should be noted that, in the image processing method provided by the embodiment of the present invention, the low-pass filter is not fixed, but dynamically changes according to the frequency domain distribution of the image, so that the problem of inapplicability to different illuminations due to the low-pass filter fixed in advance in the self-quotient graph method is solved. In addition, based on the introduction of the scale adjustment parameter α and the offset adjustment parameter β, the variation of the face texture due to the external lighting condition, the skin characteristic, and the camera parameter can be offset.
The following describes the frequency domain filtering involved in the image processing method provided by the embodiment of the present invention.
In an embodiment of the invention, the illumination estimation is performed by frequency domain filtering. The frequency domain filtering has the advantages that the problems of the size of the template and the selection of the filtering kernel in the spatial domain can be avoided, and the filtering process is easier to realize in the frequency domain. The process of frequency filtering includes:
s1001, carrying out Fourier transform on lightness characteristic data I of an image to be processed to obtain corresponding frequency spectrum distribution F g See, in particular, equation (7):
Figure GDA0003888757530000121
where s and t represent the index of I, representing the coordinate position of each pixel element in the image, and M and N are the size of the image I. F g Representing the Fourier spectrum, k and l representing F g Is used as an index of (1).
S1002, determining a low-pass filter epsilon and a Fourier spectrum F g Multiplication, as in equation (8):
|F′ g (k,l)|=|F g (k,l)|ε(k,l) (8)
wherein, | F g (k, l) | is the input Fourier spectrum F g Amplitude, | F' g (k, l) | denotes the amplitude of the filtered spectrum, where | F' g The (k, l) | is mostly low frequency information, and medium and high frequency information is almost completely eliminated by the low pass filter.
S1003, in order to obtain illumination estimation in a time domain, performing inverse Fourier transform on | F' g (k, l) | is inverse transformed, see equation (9):
Figure GDA0003888757530000122
wherein
Figure GDA0003888757530000123
Is the low frequency information of the filtered image I, i.e. the light illumination characteristic data, and 1/MN is the normalizing term of the inverse transform.
Next, a method of determining the cutoff frequency of the low-pass filter in S1002 will be described.
Different low-pass filters are selected for different images, thereby overcoming the technical drawback that a low-pass filter with a fixed cutoff frequency is not possible to fit images under all illumination.
For the selection of the cut-off frequency of the low-pass filter, the selection of the cut-off frequency of the low-pass filter is realized by calculating the energy distribution of each image in the frequency domain and then according to a proper proportional threshold value. Theoretically, the number of light sources, the direction of illumination and the influence of brightness are mainly distributed in a low-frequency part, identity information about a human face is mainly distributed in a medium-frequency part, detailed information about details on the surface of the human face is mainly distributed in a relatively stable high-frequency part, and the energy distribution situations are inseparable from the illumination situations. Therefore, by changing the energy division instead of the frequency division, the illumination condition can be accurately estimated for each image. It is noted that the first spectral distribution F g Is defined as F g The two-norm of the amplitude, as shown in equation (10):
E(F g )=||log(1+|F g |)|| 2 (10)
when the cutoff frequency is determined, an iterative method can be adopted to determine the cutoff frequency, starting from 0, the cutoff frequency of the filter is gradually increased until the energy component passed by the low-pass filter is greater than a certain proportion, namely the energy weight P. Because the energy of the illumination components is different under different illumination conditions, the value of the energy weight P is not fixed, and the values of the images under different illumination conditions are different.
For different images, energy weights P can be calculated respectively, and P changes along with the change of illumination energy of the images. Firstly, histogram equalization adjustment is carried out on an image to be processed by adopting an HE adjustment algorithm to obtain an equalized image, and further equalized brightness characteristic data of the equalized image, namely the brightness distribution of the equalized image, is obtained, so that the brightness of the image to be processed and the brightness of the equalized image are approximately in the same level, and the energy of face textures under different illumination conditions is approximately converted into a constant grade. Therefore, P can be approximated by using equation (11):
P∝E(I)/E(I eq )≈κE(I)/E(I eq ) (11)
where κ is a constant, such as a value of 0.35, 0.3, etc. E (I) is the energy of the brightness characteristic data I of the image to be processed, E (I) eq ) Is the energy of the equalized brightness feature data of the equalized image. Here, the energy means a two-norm of the image data.
In practical applications, the ideal low-pass filter is the simplest low-pass filter, and can suppress all frequencies higher than the threshold frequency (set to D0) and maintain low-frequency invariance. However, an ideal low-pass filter may generate a ringing effect, and the gray scale of the filtered time domain image changes dramatically, which seriously degrades the image quality. Therefore, more complex low-pass filters (e.g., gaussian and Butterworth filters) may be selected for use. The gaussian filter has the same size in the time and frequency domains and therefore does not cause ringing. A butterworth filter is an approximation of a gaussian filter that yields results similar to the output of a gaussian filter. However, in view of computational complexity, a butterworth filter is a better choice, whereas for a narrow range low-pass filter, a gaussian low-pass filter is a better choice. In the processing herein, a low-pass gaussian filter is employed because the illumination component is mainly located in the low-frequency region, and the gaussian filter is more appropriate.
The following describes the determination of a target image based on light illuminance characteristic data in the embodiment of the present invention.
Based on the definition of self-illumination ratio, presetting a reference gray value I of a fixed reference gray image p (such as a constant 60), the objective of illumination alignment is achieved by implementing overall minimization between the target self-quotient graph after the alignment of the formula (6) and the preset image, and the process of minimization is specifically shown in the formula (12):
argmin||I p -L R I|| F (12)
by iteratively converging the target self-quotient graph to the specified gray level image, the values of the adjustment parameters alpha and beta can be solved, so that the target self-quotient graph is determined according to the formula (6).
Example four
In the embodiment of the present invention, the method for processing an image according to the embodiment of the present invention is further described, as shown in fig. 3, including:
s301, performing color space processing on the image to be processed to obtain an HSV image corresponding to the image to be processed;
s302, extracting a V channel component of an HSV image corresponding to the image to be processed to obtain brightness characteristic data of the image to be processed;
s303, carrying out Fourier transform on lightness characteristic data of the image to be processed to obtain first frequency spectrum distribution;
s321, carrying out histogram equalization processing on the image to be processed to obtain an equalized image;
s322, performing color space processing on the balanced image to obtain an HSV image corresponding to the balanced image;
s323, extracting the V channel component of the HSV image corresponding to the balanced graph to obtain balanced lightness characteristic data;
it should be noted that the above steps S301 to S303 and S321 to S323 can be executed in parallel.
S304, determining an energy weight according to the energy of the lightness characteristic data and the energy of the balanced lightness characteristic data;
s305, determining the cutoff frequency of the low-pass filter according to the energy weight;
s306, low-pass filtering is carried out on the first frequency spectrum distribution through a low-pass filter to obtain second frequency spectrum distribution;
s307, performing Fourier inverse transformation on the second frequency spectrum distribution to obtain illumination characteristic data;
s308, determining an adjusting parameter according to the illumination characteristic data and the reference gray value;
s309, determining the target image according to the adjusting parameters.
EXAMPLE five
In this embodiment, a specific application of the image processing method provided in the embodiment of the present invention in a portrait recognition scene is illustrated by a specific image processing system.
As shown in fig. 4, the image processing system includes a portrait acquisition module 401, a portrait processing module 402, and a portrait identification module 403; the portrait acquisition module 401 is used for acquiring a portrait, the portrait processing module 402 is used for performing light removal processing on the portrait, and the portrait identification module 403 is used for portrait identification. Specifically, the method comprises the following steps:
the portrait acquisition module 401 is used for inputting and detecting a face image, and storing the portrait through a database, such as a Bosphorus database of a face. In these databases, each image must contain a human face.
In practical application, a step of face detection can be added, a face region can be segmented, and an Adaboost face detection algorithm can be adopted for detection and segmentation.
The face processing module 402 performs automatic de-illumination on the human face based on the image processing method provided by the invention, and provides an excellent data base for the face recognition module 403 before the face recognition module 403 performs face recognition. When the image is subjected to the light removal processing, the light normalization processing is performed on the face image through the formula (1). Minimizing formula (10) by using a simplex algorithm (Nelder-Meade simple algorithm), which is widely used to solve the borderless optimization problem and has simpler calculation, to obtain noise for removing the illumination field, and obtain a face de-illumination image as shown in fig. 5 or fig. 6, wherein in fig. 5, the upper four images are RGB images without de-illumination processing, and the lower four images are divided into four target images obtained after the four RGB images are subjected to the image processing provided by the embodiment of the present invention; in fig. 6, the upper four images are the gray images without the light removal processing, and the lower four images are divided into four target images obtained after the four gray images are subjected to the image processing provided by the embodiment of the invention
The image processing method provided by the embodiment of the invention is applied to an image processing system for face recognition, and has certain effectiveness in image recognition.
EXAMPLE six
In order to implement the method for processing an image provided in the foregoing embodiment, an embodiment of the present invention provides an apparatus for processing an image, as shown in fig. 7, the apparatus including: a color processing unit 701, a filtering unit 702, a determining unit 703, and a target unit 704; wherein the content of the first and second substances,
the color processing unit 701 is configured to perform color space processing on an image to be processed to obtain brightness feature data of the image to be processed;
a filtering unit 702, configured to perform low-pass filtering on the brightness feature data to obtain illumination feature data of the image to be processed;
a determining unit 703, configured to determine an adjustment parameter according to a set reference gray value and the illumination characteristic data;
and the target unit 704 is configured to perform a light removal process on the brightness feature data according to the adjustment parameter and the light feature data to obtain a target image.
In one embodiment, as shown in fig. 8, the filtering unit 702 includes: a first transformation module 7021, a filtering module 7022, and a second transformation module 7023, wherein,
a first transform module 7021, configured to perform fourier transform on the lightness feature data to obtain a first spectrum distribution corresponding to the lightness feature data;
a filtering module 7022, configured to perform filtering processing on the first spectrum distribution through a low-pass filter to obtain a second spectrum distribution;
a second transform module 7023, configured to perform inverse fourier transform on the second frequency spectrum distribution to obtain illumination characteristic data of the image to be processed.
In an embodiment, as shown in fig. 8, the filtering unit 702 further includes: a determining module 7024 configured to:
determining an energy weight according to the image to be processed;
determining a lower limit frequency point and an upper limit frequency point of the first spectrum distribution;
determining first energy of the first spectrum distribution at the lower limit frequency point and the upper limit frequency point;
determining second energy of the first frequency spectrum distribution at the lower limit frequency point and the characteristic frequency point;
and if the ratio of the first energy to the second energy is the energy weight, determining the characteristic frequency point corresponding to the second energy as the cutoff frequency of the low-pass filter.
In an embodiment, the determining module 7024, determining the energy weight according to the image to be processed includes:
carrying out histogram equalization processing on the image to be processed to obtain an equalized image;
carrying out color space processing on the balanced image to obtain balanced lightness characteristic data of the balanced image;
and determining the energy weight according to the brightness characteristic data of the image to be processed and the balanced brightness characteristic data.
In an embodiment, the tuning parameters include a scaling parameter and an offset tuning parameter; correspondingly, the determining unit 703 is specifically configured to:
determining a target self-quotient graph T according to the following formula:
Figure GDA0003888757530000181
wherein α is a scaling parameter and β is an offset adjustment parameter, beta is a reference value>
Figure GDA0003888757530000182
Is the illumination characteristic data, and I is the lightness characteristic data;
determining a two-norm between the reference gray value and the target self-quotient graph T;
and minimizing the two norms, and determining the scale adjustment parameter and the offset adjustment parameter.
In an embodiment, the tuning parameters include a scaling parameter and an offset tuning parameter; accordingly, the target unit 704 is specifically configured to:
determining a target self-quotient graph T according to the following formula:
Figure GDA0003888757530000183
wherein α is a scaling parameter and β is an offset adjustment parameter, beta is a reference value>
Figure GDA0003888757530000184
Is the illumination characteristic data, and I is the lightness characteristic data;
and taking the target self-quotient graph as the target image.
It should be noted that: the image processing apparatus provided in the above embodiment is exemplified by the division of each program module when performing image processing, and in practical applications, the processing may be distributed to different program modules according to needs, that is, the internal structure of the apparatus may be divided into different program modules to complete all or part of the processing described above. In addition, the image processing apparatus and the image processing method provided by the above embodiments belong to the same concept, and specific implementation processes thereof are described in the method embodiments in detail and are not described herein again.
EXAMPLE seven
An embodiment of the present invention further provides an image processing apparatus, including: a processor and a memory for storing a computer program operable on the processor, wherein the processor is operable when executing the computer program to perform:
carrying out color space processing on an image to be processed to obtain brightness characteristic data of the image to be processed; carrying out low-pass filtering processing on the brightness characteristic data to obtain illumination characteristic data of the image to be processed; determining an adjusting parameter through a set reference gray value and the illumination characteristic data; and performing illumination removing processing on the brightness characteristic data according to the adjustment parameters and the illumination characteristic data to obtain a target image.
The processor is further configured to, when the computer program is executed, execute the low-pass filtering processing on the brightness feature data to obtain illumination feature data of the image to be processed, where the obtaining includes:
performing Fourier transform on the lightness characteristic data to obtain a first frequency spectrum distribution corresponding to the lightness characteristic data; filtering the first frequency spectrum distribution through a low-pass filter to obtain a second frequency spectrum distribution; and carrying out Fourier inverse transformation on the second frequency spectrum distribution to obtain the illumination characteristic data of the image to be processed.
The processor is further configured to, when the computer program is executed, before the first spectral distribution is filtered by the low-pass filter to obtain a second spectral distribution, further perform: determining an energy weight according to the image to be processed; determining a lower limit frequency point and an upper limit frequency point of the first spectrum distribution; determining first energy of the first spectrum distribution at the lower limit frequency point and the upper limit frequency point; determining second energy of the first frequency spectrum distribution at the lower limit frequency point and the characteristic frequency point; and if the ratio of the first energy to the second energy is the energy weight, determining the characteristic frequency point corresponding to the second energy as the cut-off frequency of the low-pass filter.
The processor is further configured to, when the computer program is run, execute the determining the energy weight according to the image to be processed, including:
carrying out histogram equalization processing on the image to be processed to obtain an equalized image; carrying out color space processing on the balanced image to obtain balanced brightness characteristic data of the balanced image; and determining the energy weight according to the brightness characteristic data of the image to be processed and the balanced brightness characteristic data.
The processor is further configured to, when running the computer program, cause the tuning parameters to include a scaling parameter and an offset tuning parameter; correspondingly, the determining of the adjustment parameter by the set reference gray value and the illumination characteristic data comprises:
determining a target self-quotient graph T according to the following formula:
Figure GDA0003888757530000191
wherein α is a scaling parameter and β is an offset adjustment parameter, beta is a reference value>
Figure GDA0003888757530000203
Is the illumination characteristic data, and I is the lightness characteristic data; determining a binary range between the reference gray value and the target self-quotient graph TCounting; and minimizing the two norms, and determining the scale adjustment parameter and the offset adjustment parameter.
The processor is further configured to, when running the computer program, cause the tuning parameters to include a scaling parameter and an offset tuning parameter; correspondingly, the executing the illumination removing processing of the lightness feature data according to the adjusting parameters and the illumination feature data to obtain the target image comprises:
determining a target self-quotient graph T according to the following formula:
Figure GDA0003888757530000201
wherein α is a scaling parameter and β is an offset tuning parameter>
Figure GDA0003888757530000202
The lighting characteristic data is taken as the lighting characteristic data, and I is the brightness characteristic data; and taking the target self-quotient graph as the target image.
Based on this, fig. 9 is a schematic structural diagram of an image processing apparatus according to another embodiment of the present invention, and the image processing apparatus 900 shown in fig. 9 includes: at least one processor 901 and memory 902. The various components in the image processing apparatus 900 are coupled together by a bus system 903. It is understood that the bus system 903 is used to enable communications among the components.
It will be appreciated that the memory 902 can be either volatile memory or nonvolatile memory, and can include both volatile and nonvolatile memory. Among them, the nonvolatile Memory may be a Read Only Memory (ROM), a Programmable Read Only Memory (PROM), an Erasable Programmable Read-Only Memory (EPROM), an Electrically Erasable Programmable Read-Only Memory (EEPROM), a magnetic random access Memory (FRAM), a magnetic random access Memory (Flash Memory), a magnetic surface Memory, an optical Disc, or a Compact Disc Read-Only Memory (CD-ROM); the magnetic surface storage may be disk storage or tape storage. Volatile Memory can be Random Access Memory (RAM), which acts as external cache Memory. By way of illustration and not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), synchronous Static Random Access Memory (SSRAM), dynamic Random Access Memory (DRAM), synchronous Dynamic Random Access Memory (SDRAM), double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), enhanced Synchronous Dynamic Random Access Memory (ESDRAM), enhanced Synchronous Dynamic Random Access Memory (Enhanced DRAM), synchronous Dynamic Random Access Memory (SLDRAM), direct Memory (DRmb Access), and Random Access Memory (DRAM). The memory 902 described in connection with the embodiments of the invention is intended to comprise, without being limited to, these and any other suitable types of memory.
The memory 902 in the embodiment of the present invention is used to store various types of data to support the operation of the image processing apparatus 900.
The method disclosed in the above embodiments of the present invention may be applied to the processor 901, or implemented by the processor 901. The processor 901 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be implemented by integrated logic circuits of hardware or instructions in the form of software in the processor 901. The Processor 901 may be a general purpose Processor, a Digital Signal Processor (DSP), or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, or the like. Processor 901 may implement or perform the methods, steps, and logic blocks disclosed in embodiments of the present invention. A general purpose processor may be a microprocessor or any conventional processor or the like. The steps of the method disclosed by the embodiment of the invention can be directly implemented by a hardware decoding processor, or can be implemented by combining hardware and software modules in the decoding processor. The software modules may be located in a storage medium located in the memory 902, and the processor 901 reads the information in the memory 902 and performs the steps of the aforementioned methods in combination with its hardware.
In an exemplary embodiment, the image processing apparatus 900 may be implemented by one or more Application Specific Integrated Circuits (ASICs), DSPs, programmable Logic Devices (PLDs), complex Programmable Logic Devices (CPLDs), field Programmable Gate Arrays (FPGAs), general purpose processors, controllers, micro Controllers (MCUs), microprocessors (microprocessors), or other electronic components for performing the foregoing methods.
Practice of the invention
In an exemplary embodiment, the embodiment of the present invention further provides a computer readable storage medium, such as a memory 902 including a computer program, which is executable by a processor 901 of an image processing apparatus to perform the steps of the foregoing method. The computer readable storage medium may be Memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash Memory, magnetic surface Memory, optical disk, or CD-ROM.
A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, performs:
carrying out color space processing on an image to be processed to obtain lightness characteristic data of the image to be processed; carrying out low-pass filtering processing on the brightness characteristic data to obtain illumination characteristic data of the image to be processed; determining an adjusting parameter through a set reference gray value and the illumination characteristic data; and performing illumination removing processing on the brightness characteristic data according to the adjustment parameters and the illumination characteristic data to obtain a target image.
When the computer program is executed by a processor, the low-pass filtering processing on the brightness characteristic data is executed, and the obtaining of the illumination characteristic data of the image to be processed comprises:
performing Fourier transform on the lightness characteristic data to obtain a first frequency spectrum distribution corresponding to the lightness characteristic data; filtering the first frequency spectrum distribution through a low-pass filter to obtain a second frequency spectrum distribution; and carrying out Fourier inverse transformation on the second frequency spectrum distribution to obtain the illumination characteristic data of the image to be processed.
When the computer program is executed by a processor, before the first spectral distribution is filtered by a low-pass filter to obtain a second spectral distribution, the computer program further performs:
determining an energy weight according to the image to be processed; determining a lower limit frequency point and an upper limit frequency point of the first spectrum distribution; determining first energy of the first spectrum distribution at the lower limit frequency point and the upper limit frequency point; determining second energy of the first frequency spectrum distribution at the lower limit frequency point and the characteristic frequency point; and if the ratio of the first energy to the second energy is the energy weight, determining the characteristic frequency point corresponding to the second energy as the cutoff frequency of the low-pass filter.
When the computer program is executed by a processor, the determining the energy weight according to the image to be processed includes:
carrying out histogram equalization processing on the image to be processed to obtain an equalized image; carrying out color space processing on the balanced image to obtain balanced brightness characteristic data of the balanced image; and determining the energy weight according to the brightness characteristic data of the image to be processed and the balanced brightness characteristic data.
The computer program, when executed by a processor, the tuning parameters including a scaling parameter and an offset tuning parameter; correspondingly, the determining of the adjustment parameter by the set reference gray value and the illumination characteristic data comprises:
determining a target self-quotient graph T according to the following formula:
Figure GDA0003888757530000231
wherein α is a scaling parameter and β is an offset adjustment parameter, beta is a reference value>
Figure GDA0003888757530000232
Is the illumination characteristic data, and I is the lightness characteristic data; determining a two-norm between the reference gray value and the target self-quotient graph T; and minimizing the two norms, and determining the scale adjustment parameter and the offset adjustment parameter.
The computer program, when executed by a processor, the tuning parameters including a scaling parameter and an offset tuning parameter; correspondingly, the performing of the self-quotient light removal treatment on the lightness feature data according to the adjustment parameter and the illumination feature data to obtain the target image comprises:
determining a target self-quotient graph T according to the following formula:
Figure GDA0003888757530000241
wherein α is a scaling parameter and β is an offset tuning parameter>
Figure GDA0003888757530000242
The lighting characteristic data is taken as the lighting characteristic data, and I is the brightness characteristic data; and taking the target self-quotient graph as the target image.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention.

Claims (12)

1. An image processing method, characterized in that the method comprises:
carrying out color space processing on an image to be processed to obtain brightness characteristic data of the image to be processed;
performing low-pass filtering processing on the brightness characteristic data to obtain illumination characteristic data of the image to be processed, wherein the illumination characteristic data comprises illumination characteristic data obtained by removing internal illumination factors of the image to be processed and reserving external illumination factors of the image to be processed;
determining an adjustment parameter through a set reference gray value and the illumination characteristic data, wherein the reference gray value can represent a fixed reference gray image with external illumination factors removed;
and performing illumination removing processing on the brightness characteristic data according to the adjustment parameters and the illumination characteristic data to obtain a target image, wherein the brightness of the target image is aligned with that of the reference gray level image.
2. The method according to claim 1, wherein the low-pass filtering the brightness feature data to obtain the illumination feature data of the image to be processed comprises:
performing Fourier transform on the lightness characteristic data to obtain a first frequency spectrum distribution corresponding to the lightness characteristic data;
filtering the first frequency spectrum distribution through a low-pass filter to obtain a second frequency spectrum distribution;
and carrying out Fourier inverse transformation on the second frequency spectrum distribution to obtain the illumination characteristic data of the image to be processed.
3. The method of claim 2, wherein before the filtering the first spectral distribution by the low pass filter to obtain the second spectral distribution, the method further comprises:
determining an energy weight according to the image to be processed;
determining a lower limit frequency point and an upper limit frequency point of the first frequency spectrum distribution;
determining first energy of the first spectrum distribution at the lower limit frequency point and the upper limit frequency point;
determining second energy of the first frequency spectrum distribution at the lower limit frequency point and the characteristic frequency point;
and if the ratio of the first energy to the second energy is the energy weight, determining the characteristic frequency point corresponding to the second energy as the cutoff frequency of the low-pass filter.
4. The method of claim 3, wherein determining an energy weight from the image to be processed comprises:
carrying out histogram equalization processing on the image to be processed to obtain an equalized image;
carrying out color space processing on the balanced image to obtain balanced lightness characteristic data of the balanced image;
and determining the energy weight according to the brightness characteristic data of the image to be processed and the balanced brightness characteristic data.
5. The method of claim 1, wherein the tuning parameters comprise a scaling parameter and an offset tuning parameter; the determining of the adjustment parameter by the set reference gray value and the illumination characteristic data includes:
determining a target self-quotient graph T according to the following formula:
Figure FDA0003770447300000021
wherein α is a scaling parameter and β is an offset adjustment parameter, beta is a reference value>
Figure FDA0003770447300000022
Is the illumination characteristic data, and I is the lightness characteristic data;
determining a two-norm between the reference gray value and the target self-quotient graph T;
and minimizing the two norms, and determining the scale adjustment parameter and the offset adjustment parameter.
6. The method of claim 1, wherein the tuning parameters comprise a scaling parameter and an offset tuning parameter; the removing the illumination of the brightness characteristic data according to the adjustment parameter and the illumination characteristic data to obtain a target image comprises:
determining a target self-quotient graph T according to the following formula:
Figure FDA0003770447300000031
wherein α is a scaling parameter and β is an offset adjustment parameter, beta is a reference value>
Figure FDA0003770447300000032
Is the illumination characteristic data, and I is the lightness characteristic data;
and taking the image corresponding to the target self-quotient graph T as the target image.
7. An image processing apparatus, characterized in that the apparatus comprises: the device comprises a color processing unit, a filtering unit, a determining unit and a target unit; wherein the content of the first and second substances,
the color processing unit is used for carrying out color space processing on an image to be processed to obtain brightness characteristic data of the image to be processed;
the filtering unit is used for performing low-pass filtering processing on the lightness characteristic data to obtain illumination characteristic data of the image to be processed, wherein the illumination characteristic data comprises illumination characteristic data obtained by removing internal illumination factors of the image to be processed and reserving external illumination factors of the image to be processed;
the determining unit is used for determining an adjusting parameter through a set reference gray value and the illumination characteristic data, wherein the reference gray value can represent a fixed reference gray image with external illumination factors removed;
and the target unit is used for carrying out illumination removing processing on the lightness characteristic data according to the adjusting parameters and the illumination characteristic data to obtain a target image, and the brightness of the target image is aligned with that of the reference gray level image.
8. The apparatus of claim 7, wherein the filtering unit comprises: a first transformation module, a filtering module, and a second transformation module, wherein,
the first transformation module is used for performing Fourier transformation on the lightness characteristic data to obtain a first frequency spectrum distribution corresponding to the lightness characteristic data;
the filtering module is used for filtering the first frequency spectrum distribution through a low-pass filter to obtain a second frequency spectrum distribution;
and the second transformation module is used for performing inverse Fourier transformation on the second frequency spectrum distribution to obtain the illumination characteristic data of the image to be processed.
9. The apparatus of claim 8, wherein the filtering unit comprises: a determination module to:
determining an energy weight according to the image to be processed;
determining a lower limit frequency point and an upper limit frequency point of the first spectrum distribution;
determining first energy of the first spectrum distribution at the lower limit frequency point and the upper limit frequency point;
determining second energy of the first frequency spectrum distribution at the lower limit frequency point and the characteristic frequency point;
and if the ratio of the first energy to the second energy is the energy weight, determining the characteristic frequency point corresponding to the second energy as the cutoff frequency of the low-pass filter.
10. The apparatus of claim 9, wherein the determining module determines the energy weight according to the image to be processed comprises:
carrying out histogram equalization processing on the image to be processed to obtain an equalized image;
carrying out color space processing on the balanced image to obtain balanced brightness characteristic data of the balanced image;
and determining the energy weight according to the brightness characteristic data of the image to be processed and the balanced brightness characteristic data.
11. An image processing apparatus characterized by comprising: a processor and a memory for storing a computer program operable on the processor, wherein the processor is operable to perform the steps of the method of any of claims 1 to 6 when the computer program is executed.
12. A computer-readable storage medium, characterized in that a computer program is stored thereon, which computer program, when being executed by a processor, realizes the image processing method according to any one of claims 1 to 6.
CN201810621822.XA 2018-06-15 2018-06-15 Image processing method and device and computer readable storage medium Active CN110610525B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810621822.XA CN110610525B (en) 2018-06-15 2018-06-15 Image processing method and device and computer readable storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810621822.XA CN110610525B (en) 2018-06-15 2018-06-15 Image processing method and device and computer readable storage medium

Publications (2)

Publication Number Publication Date
CN110610525A CN110610525A (en) 2019-12-24
CN110610525B true CN110610525B (en) 2023-04-07

Family

ID=68888403

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810621822.XA Active CN110610525B (en) 2018-06-15 2018-06-15 Image processing method and device and computer readable storage medium

Country Status (1)

Country Link
CN (1) CN110610525B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111275804B (en) * 2020-01-17 2022-09-16 腾讯科技(深圳)有限公司 Image illumination removing method and device, storage medium and computer equipment
CN112330633B (en) * 2020-11-05 2021-06-15 哈尔滨市科佳通用机电股份有限公司 Jumper wire adhesive tape damage fault image segmentation method based on self-adaptive band-pass filtering

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101425179A (en) * 2008-11-18 2009-05-06 清华大学 Face image relighting method and device
CN101916371A (en) * 2010-09-01 2010-12-15 北京工业大学 Method for illuminating/normalizing image and method for identifying image by using same
CN102346857A (en) * 2011-09-14 2012-02-08 西安交通大学 High-precision method for simultaneously estimating face image illumination parameter and de-illumination map
CN103413275A (en) * 2013-07-26 2013-11-27 北京工业大学 Retinex night image enhancement method based on gradient zero norm minimum
WO2017152822A1 (en) * 2016-03-07 2017-09-14 华为技术有限公司 Image processing method and device

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101425179A (en) * 2008-11-18 2009-05-06 清华大学 Face image relighting method and device
CN101916371A (en) * 2010-09-01 2010-12-15 北京工业大学 Method for illuminating/normalizing image and method for identifying image by using same
CN102346857A (en) * 2011-09-14 2012-02-08 西安交通大学 High-precision method for simultaneously estimating face image illumination parameter and de-illumination map
CN103413275A (en) * 2013-07-26 2013-11-27 北京工业大学 Retinex night image enhancement method based on gradient zero norm minimum
WO2017152822A1 (en) * 2016-03-07 2017-09-14 华为技术有限公司 Image processing method and device

Also Published As

Publication number Publication date
CN110610525A (en) 2019-12-24

Similar Documents

Publication Publication Date Title
Ju et al. Single image dehazing via an improved atmospheric scattering model
Wang et al. Naturalness preserved enhancement algorithm for non-uniform illumination images
CN105957030B (en) One kind being applied to the enhancing of thermal infrared imager image detail and noise suppressing method
Zhou et al. Multi-scale retinex-based adaptive gray-scale transformation method for underwater image enhancement
Park et al. Single image haze removal with WLS-based edge-preserving smoothing filter
CN107993189B (en) Image tone dynamic adjustment method and device based on local blocking
Yang Recursive approximation of the bilateral filter
Rahman et al. Structure revealing of low-light images using wavelet transform based on fractional-order denoising and multiscale decomposition
US20210272252A1 (en) Systems and methods for image processing
CN112991197B (en) Low-illumination video enhancement method and device based on detail preservation of dark channel
CN111145086A (en) Image processing method and device and electronic equipment
CN110610525B (en) Image processing method and device and computer readable storage medium
Yu et al. Perceptually optimized enhancement of contrast and color in images
Mu et al. Low and non-uniform illumination color image enhancement using weighted guided image filtering
Kim et al. Tone mapping with contrast preservation and lightness correction in high dynamic range imaging
Panetta et al. Color image enhancement based on the discrete cosine transform coefficient histogram
Kaur et al. Image enhancement techniques: A selected review
CN116681606A (en) Underwater uneven illumination image enhancement method, system, equipment and medium
Sun et al. Dynamic range compression and detail enhancement algorithm for infrared image
Wang et al. Adaptive enhancement for nonuniform illumination images via nonlinear mapping
CN112308793A (en) Novel method for enhancing contrast and detail of non-uniform illumination image
Wang et al. A new single image dehazing method with MSRCR algorithm
Singh et al. Multiscale reflection component based weakly illuminated nighttime image enhancement
Chen et al. An adaptive image dehazing algorithm based on dark channel prior
Xi et al. Enhancement of unmanned aerial vehicle image with shadow removal based on optimized retinex algorithm

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
GR01 Patent grant
GR01 Patent grant