CN110730280B - Noise equalization method and noise removal method - Google Patents
Noise equalization method and noise removal method Download PDFInfo
- Publication number
- CN110730280B CN110730280B CN201810783704.9A CN201810783704A CN110730280B CN 110730280 B CN110730280 B CN 110730280B CN 201810783704 A CN201810783704 A CN 201810783704A CN 110730280 B CN110730280 B CN 110730280B
- Authority
- CN
- China
- Prior art keywords
- value
- values
- noise
- block
- equalized
- 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
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/80—Camera processing pipelines; Components thereof
- H04N23/81—Camera processing pipelines; Components thereof for suppressing or minimising disturbance in the image signal generation
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Image Processing (AREA)
Abstract
A noise equalization method and a noise removal method are provided. The Noise equalization method simulates the Noise intensity under different Signal intensities through a smooth curve, and then realizes the Noise equalization under different Signal intensities through an equalization curve, so that the Noise intensities under different Signal intensities are the same or approximate, namely, the Noise (Signal-Dependent Noise, SDN) related to the Signal intensity is converted into the Noise (Signal-Independent Noise, SIN) unrelated to the Signal Noise, and then the SDN characteristic of a real Signal is estimated more accurately from the pixels affected by the Noise. In addition, the noise removal method implemented by the equalization curve can simplify the calculation complexity and calculate better noise parameters, thereby improving the noise removal effect of each input pixel value in an input image and further generating an output image with lower noise.
Description
Technical Field
The present disclosure relates to a noise equalization method and a noise removal method, and more particularly, to a noise equalization method and a noise removal method based on pixel values after noise equalization.
Background
Electronic devices (such as smart phones, cameras, camcorders, etc.) generate complex noise during image acquisition. This Noise is not simply Additive White Gaussian Noise (AWGN), but is Signal-Dependent Noise (SDN), i.e., the Noise level is related to the Signal strength. Thus, different pixel values may have different intensities of noise.
For example, fig. 1 shows a schematic diagram of standard deviations of a plurality of regions of an input image. The input image Fr0 has a plurality of image blocks EN1 and EN 2. The image blocks EN1 and EN2 have a plurality of pixel values, and the overall pixel value of the image block EN1 is lower than that of the image block EN 2. As shown in fig. 1, since the input image has the SDN phenomenon, the electronic device calculates the area standard deviation Std of the pixel values in the image block EN1 to be 1.15, and calculates the area standard deviation Std of the pixel values in the image block EN2 to be 2.95, that is, the same input image has noise with different intensities.
A common Noise Estimation (Noise Estimation) is designed for gaussian white Noise and can be roughly divided into two types. One is Real-time noise estimation and the other is Offline noise estimation.
The real-time noise estimation is performed based on the image converted by a High-Pass Filter (High-Pass Filter). However, the converted image can only show the noise characteristics in the region with small variance (i.e. the smooth region in the image), and the real-time noise estimation is designed for white gaussian noise, and the image reference range of the high-pass filter also affects the accuracy of the noise estimation, so the conventional real-time noise estimation cannot estimate the SDN characteristics close to the real signal.
The off-line noise estimation is to pre-establish the SDN module to count the distribution characteristics of noise under different pixel values. The noisy input image is then used to generate a de-noised output image according to the SDN module. However, although the conventional SDN module describes the distribution characteristics of the real SDN in a linear manner, the distribution characteristics of the real SDN at different pixel values are nonlinear, and the SDN module is used to simulate the distribution characteristics of Noise at different real Signal intensities (Noise-Free Signal), when the SDN module is actually used, the SDN module can only be used to make a query according to the pixel values (Noise-Free Signal) affected by the Noise, and the real Signal (Noise-Free Signal) of the pixel cannot be obtained. Conventional off-line noise estimation also fails to accurately estimate the SDN characteristics of the real signal from the pixels affected by noise.
If the electronic device cannot estimate the SDN characteristics of the real signal, it cannot estimate better noise parameters (such as standard deviation and variance), which further affects the noise removal (noise) effect of the input image. Therefore, if the electronic device can estimate the characteristics close to the real SDN, the noise removal effect can be improved, and the quality of the input image can be improved.
Disclosure of Invention
In order to improve the noise removal effect and further improve the quality of an input image, the present disclosure provides a noise equalization method, which is suitable for an electronic device for equalizing noise under different values. The noise equalization method comprises: obtaining a plurality of representative values in a value interval and a plurality of representative standard deviations corresponding to the representative values; calculating an intermediate standard deviation corresponding to each intermediate value among the representative pixel values according to the representative values and the corresponding representative standard deviations, and mapping the representative pixel values, the representative standard deviations, the intermediate values and the intermediate standard deviations to a plurality of values and a plurality of standard deviations in a smooth curve respectively; and calculating an equalization curve according to each value in the smooth curve and the corresponding standard deviation so as to correspond each value to an equalized value. In the noise equalization method, a current value of the values plus the corresponding standard deviation is an adjusted value, and the equalization curve satisfies that the equalized value corresponding to the adjusted value minus the equalized value corresponding to the current value is 1 unit.
The scheme provides a noise removal method which is suitable for an electronic device. The noise removing method comprises the following steps: receiving an input image; obtaining a current position block in the input image, wherein the current position block corresponds to one or more input pixel values; acquiring a plurality of adjacent blocks adjacent to the current position block; mapping each input pixel value in the current position block and the adjacent blocks through an equalization curve to obtain an equalized value corresponding to each input pixel value, wherein the equalization curve comprises a plurality of values and a plurality of corresponding equalized values, a standard deviation corresponding to a current value of the values is an adjustment value, and the equalization curve satisfies that the equalized value corresponding to the adjustment value minus the equalized value corresponding to the current value is 1 unit; calculating a difference degree between each neighboring block and the current position block according to the equalized values; determining a weight value corresponding to the adjacent block according to each difference degree; and carrying out weighted average on a main pixel value in the corresponding adjacent blocks according to the weighted values to generate a corrected pixel value.
The scheme provides a noise removal method which is suitable for an electronic device. The noise removing method comprises the following steps: receiving an input image; acquiring a current position block in an input image, wherein the current position block corresponds to a plurality of input pixel values; determining a main pixel value according to the input pixel values; mapping the input pixel values through an equalization curve to obtain a plurality of corresponding equalized values, wherein the equalization curve comprises a plurality of values and a plurality of corresponding equalized values, a current value of the values plus a corresponding standard deviation is an adjustment value, and the equalization curve satisfies that the equalized value corresponding to the adjustment value minus the equalized value corresponding to the current value is 1 unit; averaging the equalized values to generate a low frequency component value; calculating a variance between the equalized values and the low-frequency component values corresponding to the input pixel values; calculating a high-frequency component value of the main pixel value in the current position block according to the low-frequency component value; calculating a high-frequency proportion of the high-frequency component value of the main pixel value according to the variance; and adding the low frequency component value to the high frequency component value of the high frequency proportion to generate an output pixel value corresponding to the main pixel value.
Drawings
FIG. 1 is a diagram illustrating standard deviations of a plurality of regions of an input image.
Fig. 2 is a schematic diagram of an electronic device according to an embodiment of the disclosure.
Fig. 3 is a flowchart illustrating a noise equalization method according to an embodiment of the present disclosure.
Fig. 4A is a detailed flowchart of step S310 according to an embodiment of the disclosure.
Fig. 4B is a schematic diagram of a test image according to an embodiment of the disclosure.
Fig. 4C is a detailed flowchart of step S310 according to another embodiment of the disclosure.
Fig. 5 shows a pixel value and a standard deviation corresponding to a dynamic range according to an embodiment of the present disclosure.
Fig. 6 is a schematic diagram of a smooth curve according to an embodiment of the present disclosure.
Fig. 7 is a diagram of an equalization curve according to an embodiment of the present disclosure.
Fig. 8 is a schematic diagram illustrating a signal distribution of an input image after equalization curve mapping according to an embodiment of the present disclosure.
Fig. 9 is a flowchart of a noise removing method according to an embodiment of the disclosure.
Fig. 10 is a schematic diagram illustrating a block of a current location and a neighboring block of an input image according to an embodiment of the present disclosure.
Fig. 11 is a flowchart illustrating a noise removing method according to another embodiment of the disclosure.
Detailed Description
Hereinafter, the present disclosure will be described in detail by illustrating various exemplary embodiments thereof with the aid of the drawings. The present concepts may, however, be embodied in different forms and should not be construed as limited to the exemplary embodiments set forth herein. Moreover, like reference numbers may indicate like elements in the drawings.
The noise equalization method provided by the embodiment of the application utilizes a smooth curve to simulate the nonlinear relation of a real SDN under different pixel values. Then, an equalization curve (composed of a plurality of pixel values and corresponding equalized values) is used to convert the noise under different pixel values into the same or similar values, i.e. the equalization curve satisfies the condition that a current pixel value is moved by a distance of a standard deviation, and the corresponding equalized values are moved by a unit. Thus, the Noise equalization method can convert the Signal-Dependent Noise (SDN) into the Signal-independent Noise (SIN), and further more accurately estimate the SDN characteristics of the true Signal (Noise-free Signal) from the pixel (Noise Signal) affected by the Noise.
In addition, the noise removal method provided in the embodiment of the present invention utilizes the equalization curve calculated by the noise equalization method to obtain SIN of each input pixel value in an input image, so as to simplify the noise removal method and calculate better noise parameters (such as standard deviation, variance, and other parameters), thereby generating an output image with lower noise.
First, fig. 2 is a schematic diagram of an electronic device 100 according to an embodiment of the disclosure. As shown in fig. 2, in the noise equalization method, the electronic device 100 equalizes the noise at different signal intensities by mapping an equalization curve (composed of a plurality of pixel values and corresponding equalized values), and further converts the noise related to the signal into the noise unrelated to the signal. In the noise removing method, the electronic device 100 obtains an equalized value corresponding to each input pixel value P0-Pn in an input image Fr1 according to the equalization curve to generate output pixel values P0 '-Pn' with lower noise. In the present embodiment, the electronic device 100 may be a smart phone, a camera, a video camera, a tablet computer, a notebook computer or other electronic devices with image capturing function, which is not limited in this application.
< one embodiment of noise equalization method >
Referring to fig. 2, the electronic device 100 includes an image acquirer 110 and an image processor 120. The image processor 120 is coupled to the image acquirer 110 and configured to perform the following steps of the noise equalization method, so as to generate a required equalization curve (composed of a plurality of pixel values and corresponding equalized values). Referring also to fig. 3, a flow chart of a noise equalization method according to an embodiment of the disclosure is shown. First, the image processor 120 of the electronic device 100 obtains a representative standard deviation corresponding to each of a plurality of representative pixel values in a pixel value interval (step S310).
In this embodiment, the pixel value interval may be designed according to a dynamic range of pixel values. For example, the dynamic range of pixel values is 0-255, which represents the possible occurrence of pixel values of 0-255. 0 represents darkest and 255 represents brightest. Therefore, the pixel value interval will be set to 0-255. The electronic device 100 further selects a plurality of suitable values from the pixel value interval as representative pixel values (for example, 5 representative pixel values, which are respectively values 0, 50, 150, 100, and 255) and calculates corresponding representative standard deviations accordingly.
More specifically, a plurality of representative pixel values and corresponding representative standard deviations can be obtained from a test image. As shown in fig. 4A and 4B, the image processor 120 first receives a test image NTC (a grayscale image is taken as an example, but not limited thereto), and the test image NTC has a plurality of different grayscale blocks Re0, Re1, Re2, Re3, Re4, Re5, Re6, Re7, Re8, Re9, Re10, and Re11 (step S410), wherein one grayscale block corresponds to a color of white, another grayscale block corresponds to a color of black, and the other grayscale blocks correspond to colors between white and black. In the present embodiment, the gray-scale block Re0 corresponds to a white color, the gray-scale block Re11 corresponds to a black color, and the other gray-scale blocks Re1-Re10 correspond to colors between the white color and the black color.
In some embodiments, after receiving the test image (step S410), the image processor 120 obtains a plurality of pixel values in each gray-scale block (step S420). In the middle of steps S410 and S420, if the test image NTC has a problem of lens shading, the image processor 120 first performs shading compensation on the test image NTC, and then performs step S420. If the NTC test image has no lens dark angle problem, the image processor 120 directly performs step S420.
In some embodiments, the edges of the gray scale blocks Re0-Re11 are susceptible to noise and form jagged blocks. In order to obtain more stable pixel values in the gray-scale blocks Re0-Re11 (e.g., similar pixel values in the corresponding gray-scale blocks, in other words, obtain pixel values in the flat areas of the corresponding gray-scale blocks), the image processor 120 may also interpolate each gray-scale block Re0-Re11 by a predetermined distance (e.g., 2 pixel distances) to form a target block, and then obtain the pixel values in each target block. Taking the gray-scale block Re0 as an example, the image processor 120 obtains each pixel value within a target block Fd0 smaller than the gray-scale block Re0 as more stable pixel values in the gray-scale block Re 0. Of course, the image processor 120 may also obtain each pixel value or pixel values with similar values in each gray-scale block Re0-Re11, which is not limited in this application.
After obtaining the pixel values of each gray-scale block Re0-Re11 (step S420), the image processor 120 calculates the representative pixel value and the representative standard deviation of the corresponding gray-scale block from the obtained pixel values in each gray-scale block Re0-Re11 (step S430). Taking the gray-scale block Re0 as an example, if the image processor 120 obtains 81 pixel values, the image processor 120 averages 81 pixel values to generate an average pixel value as the representative pixel value I0 of the gray-scale block Re0, and calculates a standard deviation of the 81 pixel values as the representative standard deviation σ 0 of the gray-scale block Re0 according to the average pixel value. The representative pixel values and the representative standard deviations corresponding to the gray-scale blocks Re0-Re11 are summarized in the following table.
TABLE < I >
| Gray-scale block | (representing pixel value, representing standard deviation) | Gray-scale block | (representing pixel value, representing standard deviation) |
| Re0 | (I0,σ0)＝(0,0.36) | Re6 | (I6,σ6)＝(120,4.70) |
| Re1 | (I1,σ1)＝(4,0.96) | Re7 | (I7,σ7)＝(150,5.26) |
| Re2 | (I2,σ2)＝(9,1.36) | Re8 | (I8,σ8)＝(200,6.07) |
| Re3 | (I3,σ3)＝(30,2.37) | Re9 | (I9,σ9)＝(240,6.59) |
| Re4 | (I4,σ4)＝(50,3.07) | Re10 | (I10,σ10)＝(250,5.17) |
| Re5 | (I5,σ5)＝(100,4.30) | Re11 | (I11,σ11)＝(255,0) |
In other embodiments of step S310 of fig. 3, the image processor 120 may also obtain a plurality of representative pixel values and corresponding representative standard deviations through a plurality of test images NTC. As shown in FIG. 4C, the image processor 120 first receives a test image having a plurality of different gray-scale blocks at a plurality of time points (step S460). For example, the image processor 120 acquires the test image NTC at 80 different time points, and the test image NTC has a plurality of different gray-scale blocks Re0-Re 11.
Then, the image processor 120 obtains at least one pixel value in each gray-scale block of each test image (step S470). The embodiment of the image processor 120 obtaining a plurality of pixel values in the gray-scale blocks can be substantially derived from step S420 of fig. 4A, and therefore, the description thereof is omitted here. In addition, since there are multiple test images at different time points, in some embodiments, one pixel value may be obtained in each gray-scale block of each test image.
Then, the image processor 120 calculates a block pixel value and a block standard deviation of each gray-scale block according to the obtained pixel values in each test image (step S480). The implementation of the image processor 120 calculating the block pixel value and the block standard deviation of each gray-scale block in each comparison image is substantially the same as the implementation of the representative pixel value (corresponding to the block pixel value) and the representative standard deviation (corresponding to the block standard deviation) in step S430 of fig. 4A, and therefore, the description thereof is omitted here.
Finally, the image processor 120 averages the block pixel values of the gray-scale blocks at the same position in each test image as the corresponding representative pixel values, and averages the block standard deviations of the gray-scale blocks at the same position as the corresponding representative standard deviations (step S490). Taking the gray-scale block Re0 of the test image NTC as an example, the image processor 120 uses the block pixel value of the gray-scale block Re0 at the same position as the corresponding representative pixel value I0 and the block standard deviation of the gray-scale block Re0 at the same position as the corresponding representative standard deviation σ 0 in each test image NTC. The representative pixel values and the representative standard deviations corresponding to the gray-scale blocks Re0-Re11 are shown in the table.
As shown in fig. 5, in the present embodiment, the pixel value interval is set to be the same as the dynamic range, i.e., 0 to 255. The image processor 120 corresponds the representative pixel values I0-I11 of the gray-scale blocks Re0-Re11 and the corresponding representative standard deviations σ 0- σ 11 to the partial pixel values in the pixel value interval and the corresponding standard deviations.
Referring to fig. 3 and fig. 5 to 6, in order to find the standard deviation corresponding to each pixel value in the pixel value interval, the image processor 120 calculates the intermediate standard deviations corresponding to a plurality of intermediate pixel values between the representative pixel values I0-I11 to form a smooth curve Crv 1. More specifically, the image processor 120 calculates an intermediate standard deviation corresponding to each intermediate pixel value between the representative pixel values according to the representative pixel values and the corresponding representative standard deviations, and maps the representative pixel values, the corresponding representative standard deviations, the intermediate pixel values and the intermediate standard deviations to a plurality of pixel values and a plurality of standard deviations in the smooth curve Crv1, respectively, thereby finding a standard deviation corresponding to each pixel value in the pixel value interval (step S320).
It is noted that two adjacent pixel values have the characteristic that the noise magnitude is close. Therefore, the standard deviation corresponding to each pixel value in the smoothing curve Crv1 is better to be closer to the representative standard deviations σ 0 to σ 11 corresponding to the representative pixel values I0 to I11 (hereinafter referred to as data fit), and the variation between the standard deviations corresponding to each pixel value is better to be smooth (hereinafter referred to as smoothness). Therefore, in the present embodiment, the electronic device 100 estimates the intermediate standard deviation corresponding to each intermediate pixel value through a smooth curve function. The smooth curve function of this embodiment is as equation (1).
Where n is the number of representative pixel values (the number of representative pixel values I0-I11 is 12 in this embodiment), m is the number of pixel values in the pixel value interval (the number of pixel values is 256 in this embodiment), λ is a control parameter for the degree of smoothing, x is a standard deviation vector describing the corresponding standard deviation of each signal in the dynamic range (the standard deviations corresponding to the pixel values 0-255 in this embodiment), and A is1To describe a first matrix of each representative pixel value (I0-I11 in this embodiment), b1 is a first vector of each representative standard deviation (σ 0- σ 11 in this embodiment), A2The second vector b2 is a 0 vector for describing a second matrix in which each standard deviation corresponds to a smooth relationship between the previous standard deviation and the next standard deviation.
Further, the first momentArray A1Corresponding first vector b1 is used to describe the data. Therefore, in this embodiment, the first matrix A1And the first vector b1 are shown as formula (2) and formula (3), respectively.
In equation (2), when the ith sampled pixel value belongs to the jth pixel value in the pixel interval, V1(i, j) ═ 1; other cases, V1(i, j) ═ 0. In equation (3), n is the number of representative pixel values, and σ 0- σ n-1 is a numerical value representing the standard deviation. A first matrix A1The matrix size of (number of representative pixel values x number of pixel values within the pixel interval) and the size of the first vector b1 is (number of representative pixel values x 1). However, the first matrix A1The first vector b1 may be designed in other ways as described above, but is not limited thereto.
In addition, a second matrix A2(indicating that the previous standard deviation and the next standard deviation corresponding to each standard deviation are in a smooth relationship) and the corresponding second vector b2 are used to describe data smoothing. Therefore, in this embodiment, the second matrix A2And the second vector can be shown as the following equations (4) and (5), respectively.
Wherein, V2(i, j) is a second matrix A2M is the number of pixel values in the pixel value interval. So that the second matrix A2The matrix size of (number of pixel values in the pixel interval x number of pixel values in the pixel interval) and the second vector b2 is (number of pixel values in the pixel interval x 1). And a second matrix A2The second vector b2 may be designed in other ways as described above, but is not limited thereto.
Accordingly, the image processor 120 estimates each pixel value (0-255 in this embodiment) and the corresponding standard deviation (i.e., vector x) in the pixel interval according to the smooth curve function of equation (1) to form a smooth curve Crv1, as shown in fig. 6. At this time, the smooth curve Crv1 represents the nonlinear relationship of the simulated real SDN at different pixel values.
Referring to fig. 3 and 7, in order to convert the signal-dependent noise (SDN) into the signal-independent noise (SIN), the image processor 120 calculates an equalization curve Crv2 according to each pixel value and the corresponding standard deviation in the smoothing curve Crv1, so as to correspond each pixel value to an equalized value (step S330). More specifically, the image processor 120 converts the noise at different pixel values into the same or similar values through the equalization curve Crv 2. Therefore, the equalization curve Crv2 needs to satisfy the requirement that a current pixel value is shifted by a standard deviation distance, and the corresponding equalized value is shifted by one unit.
Accordingly, in the present embodiment, the equalization curve Crv2 needs to satisfy that the equalized value corresponding to an adjusted pixel value minus the equalized value corresponding to the current pixel value is 1 unit, where the adjusted pixel value represents a current pixel value plus the corresponding standard deviation. The satisfying condition can be expressed by the following formula (6).
Teq (a + Xa) -Teq (a) ═ 1, 0. ltoreq. a.ltoreq.m-1 formula (6)
Where Teq is the equalized value, a is a current pixel value in the equalization curve Crv2, Xa is a standard deviation corresponding to the current pixel value, and m is the number of pixel values in the pixel interval.
More specifically, the equalization curve Crv2 also satisfies that the equalized value corresponding to the next pixel value and the equalized value corresponding to the previous pixel value are subtracted and then divided by 2 (which can be regarded as the slope of the curve at the current pixel value a) to be the reciprocal of the standard deviation corresponding to the current pixel value. The satisfying condition can be expressed by the following formula (7).
In addition, when a is equal to 0, Teq (1) -Teq (0) is equal to 1/X0; when a is m-1, Teq (m-1) -Teq (m-2) is 1/Xm-1, where Teq is a post-equalization value, a is a current pixel value in the equalization curve Crv2, Xa is a standard deviation corresponding to the current pixel value, and m is the number of signals in the dynamic range.
Therefore, in the present embodiment, the electronic device 100 calculates an equalized value corresponding to each pixel value by equation (7), and arranges the equalized value by equation (8).
The image processor 120 performs matrix operation on the equation (8) to obtain equalized values Teq (0), Teq (1) …, Teq (255) corresponding to each pixel value (i.e. 0-255), so that the equalized curve Crv2 can satisfy a condition that a current pixel value moves by a distance of one standard deviation and the corresponding equalized value moves by one unit, and further, noise (SDN) related to a signal is converted into noise (SIN) unrelated to the signal.
For example, the image processor 120 receives a current pixel value of 30. Referring to fig. 6-7, the electronic device 100 obtains, according to the smooth curve Crv1, that the standard deviation of the current pixel value 30 is 2.37 (as shown in fig. 6 and table < one >), the equalized value Teq (31) corresponding to the next pixel value 31 is 22.31, and the equalized value Teq (29) corresponding to the previous pixel value 29 is 21.47. The image processor 120 calculates [ Teq (31) -Teq (29) ]/2 ═ 0.42 (22.31-21.47)/2 and 1/2.37 ═ 0.42 from equation (7), that is, the equalization curve Crv2 satisfies the condition of equation (7).
Accordingly, as shown in fig. 8, after receiving each input pixel value P0-Pn of the input image Fr1 of fig. 2, the image processor 120 obtains an equalized value corresponding to each input pixel value P0-Pn through the equalization curve Crv2, thereby generating an equalized input image Fr 1' such that the noise values at different pixel values are the same or close to each other. As shown in fig. 8, in the equalized input video Fr1 ', the video processor 120 calculates the regional standard deviation Std of the pixel values in the video block ZN1 to be 4.32, and calculates the regional standard deviation Std of the pixel values in the video block ZN2 to be 4.33, that is, the equalized input video Fr 1' has noise with the same or close intensity. Therefore, the image processor 120 can have better noise removing effect for removing noise from the equalized input image Fr 1'.
After obtaining the equalization curve, the image processor 120 stores the equalization curve in a memory (not shown) in the electronic device 100 or an external memory (not shown). When the image processor 120 receives each input pixel value of an input image, it can obtain an equalized value corresponding to each input pixel value according to the equalization curve, and generate a lower-noise output pixel value P0 '-Pn' by a noise removal method.
< one embodiment of noise removing method >
Referring to fig. 2 and 9, fig. 9 is a flowchart illustrating a noise removing method according to an embodiment of the disclosure. First, the image processor 120 of the electronic device 100 receives an input pixel value of each pixel position in an input image (step S910). For example, as shown in fig. 10, the input image Fr has a size of 8 × 8 and 64 input pixel values. The image processor 120 receives the input pixel values P0-P63 for each pixel position in the input image Fr.
Next, the image processor 120 sequentially obtains a current position block from the pixel positions of the input image (step S920). The current location block may correspond to one input pixel value or a plurality of input pixel values. Furthermore, the shape of the current location block may be square. The size of the input image is M, and the size of the current position block is K, wherein 1< ═ K < M, and K and M are positive integers. Therefore, the image processor 120 sequentially obtains the corresponding input pixel values of the current position block according to the shape and size of the current position block. For example, if the size of the current position block is 1 × 1, the image processor 120 sequentially obtains the corresponding input pixel values in units of 1 pixel position. For another example, if the size of the current position block is 3 × 3, the image processor 120 sequentially obtains the corresponding input pixel values in units of 9 pixel positions.
In other embodiments, the current location block may have other shapes. The image processor 120 may have repeated input pixel values each time it acquires the current position block, which is not limited in this disclosure.
After obtaining the current position block (step S920), the image processor 120 obtains a plurality of neighboring blocks neighboring the current position block (step S930). The shape and size of the current location block is the same as the shape and size of the neighboring block. More specifically, during the process of acquiring the neighboring block, the image processor 120 determines that the block at the current position has one input pixel value or a plurality of input pixel values. If the image processor 120 determines that the current position block has only one input pixel value, the image processor 120 obtains the input pixel values of the blocks adjacent to the current position according to the number of the adjacent blocks, and uses the obtained input pixel values as the adjacent blocks.
If the image processor 120 determines that the current position block has a plurality of input pixel values, the image processor 120 searches for a block with the same size in the neighboring area according to the size of the current position area to generate a corresponding neighboring block.
The current position block Pch corresponds to 9 input pixel values P26-P28, P34-P36, P42-P44 and the number of neighboring blocks is 2 for illustration. As shown in FIG. 10, the image processor 120 will generate the corresponding neighboring blocks Qch1 and Qch2 by diffusing outward the distance of 1 pixel position based on the input pixel values P26 and P44. The neighboring block Qch1 corresponds to 9 input pixel values P17-P19, P25-P27, P33-P35. The neighboring block Qch2 corresponds to 9 input pixel values P35-P37, P43-P45, P51-P53. At this time, the current position block Pch and the neighboring blocks Qch1 and Qch2 have the same shape and size.
Referring back to fig. 9, after obtaining a plurality of neighboring blocks (step S930), the image processor 120 then maps the current position block and each of the input pixel values in the neighboring blocks via an equalization curve to obtain a certain equalized value corresponding to each of the input pixel values (step S940). In the present embodiment, referring to fig. 7, the image processor 120 maps each of the input pixel values P17-P19, P25-P28, P33-P37, P42-P45, and P51-P53 of the current position block Pch and the neighboring blocks Qch1-Qch2 via an equalization curve Crv2 to obtain corresponding equalized values Teq (P17) -Teq (P19), Teq (P25) -Teq (P28), Teq (P33) -Teq (P37), Teq (P42) -Teq (P45), and Teq (P51) -Teq (P53).
Next, the image processor 120 calculates a difference between each neighboring block and the current position block (step S950). More specifically, the image processor 120 calculates the difference between the equalized values of the signals in the current block and the equalized values of the signals in each neighboring block to be several standard deviations, so as to generate the difference degree of each neighboring block. In connection with the above example, the image processor 120 calculates the difference between the equalized values in the current position block Pch and the equalized values in the neighboring blocks Qch1 as several standard deviations to generate the difference Diff (Qch1) of the neighboring blocks Qch 1. The image processor 120 also calculates the differences between the equalized values in the current position block Pch and the equalized values in the neighboring blocks Qch2 as several standard deviations, corresponding to the difference Diff (Qch2) of the neighboring block Qch 2.
In this embodiment, the difference degree of the neighboring blocks is calculated by a difference degree function, and is shown in the following formula (9).
Wherein S is the number of pixels in the current position block Pch, Teq (Pch) is the equalized value of the current position block, Teq (Qchn) is the equalized value of the nth neighboring block, | Teq (Pch) -Teq (Qchn) | is the difference between several standard deviations between the equalized value of the current position block and the equalized value of the nth neighboring block, and Diff (Qchn) is the difference between the nth neighboring block and the current position block Pch.
The difference Diff (Qch1) of the neighboring blocks Qch1 calculated by the image processor 120 is taken as an example. The number S of pixels in the current position block Pch is 9. The total of | Teq (Pch) -Teq (Qchn) | is | Teq (P26) -Teq (P17) | + | Teq (P27) -Teq (P18) | + | Teq (P28) -Teq (P19) | + | Teq (P34) -Teq (P25) | + | Teq (P35) -Teq (P26) | + | Teq (P36) -Teq (P27) | + | Teq (P42) -Teq (P33) | + | Teq (P43) -Teq (P34) | + | Teq (P44) -Teq (P35) |. Thus, the image processor 120 can calculate the difference Diff of the neighboring block Qch1 by dividing the total of | Teq (Pch) -Teq (Qchn) | by the number S (Qch 1).
After obtaining the difference degree of each neighboring block, the image processor 120 determines a weight value of the corresponding neighboring block according to each difference degree (step S960). More specifically, the smaller the difference, the more similar the representative neighboring block Qch1 and the current location block Pch are. At this time, the image processor 120 determines a larger weight value according to the degree of difference. On the contrary, if the difference degree is larger, the neighboring block Qch1 and the current position block Pch are less similar. At this time, the image processor 120 determines a smaller weight value according to the degree of difference.
In the present embodiment, the image processor 120 calculates the weight value through a weight value function, as shown in the following formula (10).
Where ω (Qchn) is the weight value of the nth neighboring block, diff (Qchn) is the degree of difference of the nth neighboring block, k is a reference constant for similarity and dissimilarity, and k is the standard deviation in physical sense.
For example, the image processor 120 determines the weight value ω (Qch1) of the neighboring block Qch1 and the weight value ω (Qch2) of the neighboring block Qch 2. In this example, k is 2, the degree of difference Diff (Qch1) is 2, and the degree of difference Diff (Qch2) is 3. Therefore, the weight value ω (Qch1) ═ exp (-2/4) ═ 0.6 for the neighboring block Qch 1. The weight value ω (Qch2) ═ exp (-7/4) ═ 0.17 of the neighboring block Qch 2. Accordingly, as can be seen from the above example, when the degree of difference between the neighboring tile Qch1 and the current position tile Pch is smaller, it indicates that the neighboring tile Qch1 is more similar to the current position tile Pch, and the image processor 120 determines a larger weight, and the larger weight represents a higher reference value.
After determining the weight values of each neighboring block (step S960), the image processor 120 performs a weighted average of the main pixel values of the corresponding neighboring blocks according to the weight values to generate a modified pixel value (step S970). In this embodiment, the main pixel value in the neighboring block may be the input pixel value at a middle pixel position, or the average value of each input pixel value in the neighboring block. Of course, the main pixel values in the adjacent blocks can be obtained by other methods, which is not limited in this disclosure.
In addition, the image processor 120 of the present embodiment calculates the corrected pixel value by a correction function, which is shown in the following formula (11).
Where COR is the corrected pixel value, ω (Qchn) is the weight value of the nth neighboring block, and m (Qchn) is a main pixel value in the nth neighboring block.
Taking advantage of the above example, the weight value ω (Qch1) of the neighboring block Qch1 is 0.6, and the weight value ω (Qch2) of the neighboring block Qch2 is 0.17. The main pixel value of the neighboring block Qch1 is, for example, the input pixel value P26 of an intermediate pixel position, and the main pixel value of the neighboring block Qch1 is, for example, the input pixel value P44 of an intermediate pixel position. Therefore, the corrected pixel value COR ═ [ (0.6 × P26) + (0.17 × P44) ]/(0.6+ 0.17).
In some embodiments, the current location block Pch itself can also be regarded as one of the neighboring blocks, and its corresponding weight value can be set by itself, such as 0.8, so that the corrected pixel value COR [ (0.8 × P35) + (0.6 × P26) + (0.17 × P44) ]/(0.8+0.6+ 0.17).
Next, the image processor 120 generates each output pixel value of the current position block by using the modified pixel values (step S980). In some embodiments, if the modified pixel value is generated only for a specific pixel of the current position block, the step S980 can be omitted, but the disclosure is not limited thereto.
Therefore, when the image processor 120 performs the above-mentioned noise removal method (i.e., steps S910-S970) on each of the input pixel values P0-Pn of the input image Fr1, the image processor 120 will generate the output pixel values P0 '-Pn' with lower noise.
< Another embodiment of the noise removing method >
Referring to fig. 2 and 11, fig. 11 is a flowchart illustrating a noise removing method according to another embodiment of the disclosure. First, the image processor 120 of the electronic device 100 receives an input pixel value of each pixel position in an input image (step S1110). For example, as shown in fig. 10, the input image Fr has a size of 8 × 8 and 64 input pixel values. The image processor 120 receives the input pixel values P0-P63 for each pixel position in the input image Fr.
Next, the image processor 120 sequentially obtains a current position block from the pixel positions of the input image (step S1120). The current position block corresponds to a plurality of input pixel values, and the input pixel values have a main pixel value. More specifically, the current location block is square in shape. The size of the input image is M, and the size of the current position block is K, wherein 1< K < M, and K and M are positive integers. Therefore, the image processor 120 sequentially obtains the corresponding input pixel values of the current position block according to the shape and size of the current position block. For example, if the size of the current position block is 3 × 3, the image processor 120 sequentially obtains the corresponding input pixel values in units of 9 pixel positions.
In other embodiments, the current location block may have other shapes. The image processor 120 may have repeated input pixel values each time it acquires the current position block, which is not limited in this disclosure. In addition, in this embodiment, the main pixel value may be an input pixel value located at an intermediate pixel position in the corresponding current position block, or an average value of each input pixel value in the current position block. Of course, the main pixel value in the current position block can be obtained by other methods, which is not limited in this disclosure.
After obtaining the current position block (step S1120), the image processor 120 then maps each input pixel value in the current position block via the equalization curve to obtain a certain equalized value corresponding to each input pixel value (step S1130). Please refer to fig. 7 and fig. 10, which take the current location block Pch as an example for explanation. The image processor 120 maps each of the input pixel values P26-P28, P34-P36, P42-P44 in the current position block Pch via an equalization curve Crv2 to obtain corresponding equalized values Teq (P26) -Teq (P28), Teq (P34) -Teq (P36), Teq (P42) -Teq (P44).
Next, the image processor 120 averages the equalized values corresponding to the input pixel values in the current position block to generate a low frequency component value (step S1140). In connection with the above example, the low frequency component value E [ Teq (Pch)) ] [ (Teq (P26) + Teq (P27) + Teq (P28) + Teq (P34) + Teq (P35) + Teq (P36) + Teq (P42) + Teq (P43) + Teq (P44) ]/9 of the current location block Pch.
Then, the image processor 120 calculates a variance between the equalized value and the low frequency component value corresponding to the input pixel value in the current position block (step S1150). More specifically, the image processor 120 calculates the variance through a variance function, as shown in the following formula (12).
Where var (pch) is the variance in the current position block, teq (x) is the equalized value corresponding to one input pixel value, E [ teq (pch) ] is the low frequency component value in the current position block, and P is the number of these input pixel values in the current position block.
Taking advantage of the above example, the variance of the current location block Pch, var (Pch) ((Teq [ P26] -E [ Teq (Pch) ]) 2+ (Teq [ P27] -E [ Teq (Pch)) ]) 2+ (Teq [ P28] -E [ Teq (Pch)) ]) 2+ (Teq [ P34] -E [ Teq (Pch)) ]) 2+ (Teq [ P35] -E [ Teq (Pch)) ]) 2+ (Teq [ P36] -E [ Teq (Pch)) ]) 2+ (Teq [ P42] -E [ Teq (Pch) ]) 2+ (Teq [ P43] -E [ Teq (Pch)) (Teq [ P (44) -E [ Teq ]/2 ]/9).
Then, the image processor 120 calculates a high frequency component value of the main pixel values in the current position block according to the low frequency component value (step S1160). More specifically, image processor 120 subtracts the low frequency component values from the main pixel values to produce high frequency component values. Taking the above example as an example, the main pixel value in the current position block Pch is the input pixel value P35. Therefore, the high-frequency component value H (teq (pch)) — the input pixel value P35 — the low-frequency component value E [ teq (pch)) ].
It should be noted that the main pixel values of the current location block Pch consist of low frequency component values E [ teq (Pch)) ] and high frequency component values H (teq (Pch)). The low frequency component value E [ teq (pch) ] describes the flat portion in the main pixel value. The high frequency component value H (teq (pch)) describes the detail and texture portion in the main pixel value. However, noise (noise) in the main pixel value is also likely to be present in the high-frequency component value H (teq (pch)).
Therefore, the image processor 120 calculates a high frequency ratio of the high frequency component values of the main pixel values in the current position block according to the variance to obtain the noise-removed high frequency component values (step S1170). In the present embodiment, the high frequency ratio of the high frequency component values is calculated by a high frequency ratio function, and is shown in the following formula (13).
Wherein RTO (pch) is the high frequency ratio of the high frequency component values in the current location block, Var (pch) is the variance in the current location block, and "1" represents a noise distribution constant.
Thus, if the difference between the variance and the noise distribution constant is smaller, it means that more noise exists in the high frequency component value H (teq (pch)). At this time, the image processor 120 calculates a lower high frequency rate RTO (pch) to reduce the high frequency component value H (teq (pch)). On the contrary, if the difference between the variance and the noise distribution constant is larger, it means that more details and textures exist in the high frequency component value H (teq (pch)). At this time, the image processor 120 calculates a higher high frequency rate RTO (pch) to increase the high frequency component value H (teq (pch)).
Finally, the image processor 120 adds the low frequency component value to the high frequency component value of the high frequency ratio in the current position block to generate an output pixel value corresponding to the main pixel value (step S1180). Taking the above example as an example, the main pixel value in the current position block Pch is taken as the input pixel value P35. The image processor 120 adds the low frequency component value E [ Teq (pch)) ] to the high frequency component value H (Teq (pch)) of the high frequency scale RTO (pch) to generate an output pixel value P35' of the input pixel value P35.
Therefore, when the image processor 120 performs the above-mentioned noise removal method (i.e., steps S1110-S1180) on each of the input pixel values P0-Pn of the input image Fr1, the image processor 120 will generate the output pixel values P0 '-Pn' with lower noise.
In summary, embodiments of the present disclosure provide a noise equalization method and a noise removal method for different pixel values. The noise equalization method simulates the noise degree under different pixel values through a smooth curve, and then realizes the noise equalization degree under different pixel values through an equalization curve, so that the noise values under different pixel values are the same or approximate, namely the noise related to the signal is converted into the noise unrelated to the signal, and the SDN characteristic of the real signal is estimated more accurately from the signal affected by the noise. In addition, the noise removal method of the scheme can simplify the calculation complexity and calculate better noise parameters by utilizing the equalization curve, not only has consistent noise removal capability on pixel values under different signal intensities, but also can improve the noise removal effect of each input pixel value in an input image so as to generate an output pixel value with lower noise.
Claims (10)
1. A noise equalization method is applied to an electronic device for equalizing noise under different values, and comprises the following steps:
obtaining a plurality of representative values in a value interval and a plurality of representative standard deviations corresponding to the representative values;
calculating an intermediate standard deviation corresponding to each intermediate value among the representative values according to the representative values and the corresponding representative standard deviations, and respectively mapping the representative values, the representative standard deviations, the intermediate values and the intermediate standard deviations to a plurality of values and a plurality of standard deviations in a smooth curve; and
calculating an equalization curve according to each value in the smooth curve and the corresponding standard deviation so as to correspond each value to an equalized value;
wherein, a current value of the values plus the corresponding standard deviation is an adjustment value, and the equalization curve satisfies that the equalized value corresponding to the adjustment value minus the equalized value corresponding to the current value is 1 unit.
2. The method of claim 1, wherein a slope of the current value in the equalization curve is an inverse of the standard deviation corresponding to the current value.
3. The method of claim 1, wherein the step of obtaining the representative values and the corresponding representative standard deviations further comprises:
receiving a test image, wherein the test image has a plurality of different blocks;
respectively obtaining a plurality of first values in each block; and
in each block, the representative value and the representative standard deviation corresponding to the block are calculated according to the acquired first values so as to obtain the representative values and the representative standard deviations.
4. The method of claim 1, wherein the step of obtaining the representative values and the corresponding representative standard deviations further comprises:
receiving a test image at a plurality of time points, wherein the test image has a plurality of different blocks;
respectively obtaining at least one first numerical value in each block of each test image;
in each test image, calculating a block value and a block standard deviation of each block according to the acquired first values; and
and calculating the corresponding representative value according to the block values of the blocks at the same position, and taking the block standard deviations of the blocks at the same position as the corresponding representative standard deviations.
5. A noise removing method is suitable for an electronic device and comprises the following steps:
receiving an input image;
obtaining a current position block in the input image, wherein the current position block corresponds to one or more input pixel values;
acquiring a plurality of adjacent blocks adjacent to the current position block;
mapping each input pixel value in the current position block and the adjacent blocks through an equalization curve to obtain an equalized value corresponding to each input pixel value, wherein the equalization curve comprises a plurality of values and a plurality of corresponding equalized values, a current value of the values plus a corresponding standard deviation is an adjusted value, and the equalization curve satisfies that the equalized value corresponding to the adjusted value minus the equalized value corresponding to the current value is 1 unit;
calculating a difference degree between each neighboring block and the current position block according to the equalized values;
determining a weight value corresponding to the neighboring block according to each of the differences;
and carrying out weighted average on a main pixel value in the corresponding adjacent blocks according to the weighted values to generate a corrected pixel value.
6. The method of claim 5, wherein the step of performing a weighted average of the main pixel values in the neighboring blocks according to the weighting values to generate the modified pixel value further comprises:
setting a current weight value of the current position block;
and performing weighted average on the main pixel values in the adjacent blocks and a main pixel of the current position block according to the weight values and the current weight value to generate the corrected pixel value.
7. The method of claim 5, wherein the step of determining the weight value of each neighboring block further comprises:
calculating a standard deviation error between the equalized values in the current position block and the equalized values in each of the neighboring blocks to correspondingly generate the difference degree of each of the neighboring blocks, wherein the weight value corresponding to the difference degree is larger if the difference degree is smaller, and the weight value corresponding to the difference degree is smaller if the difference degree is larger.
8. The method as claimed in claim 5, wherein the main pixel value in each of the neighboring blocks is the input pixel value located at a middle pixel position of each of the neighboring blocks or an average value of each of the input pixel values in each of the neighboring blocks.
9. A noise removing method is suitable for an electronic device and comprises the following steps:
receiving an input image;
obtaining a current position block in the input image, wherein the current position block corresponds to a plurality of input pixel values;
determining a main pixel value according to the input pixel values;
mapping the input pixel values through an equalization curve to obtain a plurality of corresponding equalized values, wherein the equalization curve comprises a plurality of values and a plurality of corresponding equalized values, a current value of the values plus a corresponding standard deviation is an adjustment value, and the equalization curve satisfies that the equalized value corresponding to the adjustment value minus the equalized value corresponding to the current value is 1 unit;
averaging the equalized values to generate a low frequency component value;
calculating a variance between the equalized values and the low frequency component values corresponding to the input pixel values;
calculating a high frequency component value of the main pixel value in the current position block according to the low frequency component value;
calculating a high frequency proportion of the high frequency component value of the main pixel value according to the variance; and
the low frequency component value is added to the high frequency component value of the high frequency proportion to generate an output pixel value corresponding to the main pixel value.
10. The method of claim 9, wherein the main pixel value is one of the input pixel values or an average of the input pixel values.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810783704.9A CN110730280B (en) | 2018-07-17 | 2018-07-17 | Noise equalization method and noise removal method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810783704.9A CN110730280B (en) | 2018-07-17 | 2018-07-17 | Noise equalization method and noise removal method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110730280A CN110730280A (en) | 2020-01-24 |
| CN110730280B true CN110730280B (en) | 2021-08-31 |
Family
ID=69217413
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810783704.9A Active CN110730280B (en) | 2018-07-17 | 2018-07-17 | Noise equalization method and noise removal method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110730280B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5923775A (en) * | 1996-04-04 | 1999-07-13 | Eastman Kodak Company | Apparatus and method for signal dependent noise estimation and reduction in digital images |
| EP1309185A1 (en) * | 2001-10-31 | 2003-05-07 | Sony International (Europe) GmbH | Image noise detection |
| JP2008131529A (en) * | 2006-11-24 | 2008-06-05 | Sony Corp | Imaging apparatus, and image processing apparatus, method and program |
| CN102609914A (en) * | 2012-01-17 | 2012-07-25 | 天津大学 | Signal-correlated noise estimating method for image sensor |
| CN102801943A (en) * | 2011-05-24 | 2012-11-28 | 联发科技股份有限公司 | Methods for performing adaptive equalization and associated apparatus |
| CN102968770A (en) * | 2012-11-30 | 2013-03-13 | 华为技术有限公司 | Method and device for eliminating noise |
| CN103177423A (en) * | 2011-10-07 | 2013-06-26 | 伊姆普斯封闭式股份有限公司 | Method of noise reduction in digital x-ray frames series |
| CN107993200A (en) * | 2017-11-02 | 2018-05-04 | 天津大学 | Picture noise level estimation method based on deep learning |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8478062B2 (en) * | 2009-10-28 | 2013-07-02 | Apple Inc. | Reducing signal-dependent noise in digital cameras |
| US8977049B2 (en) * | 2010-01-08 | 2015-03-10 | Nvidia Corporation | System and method for estimating signal-dependent noise of an image |
-
2018
- 2018-07-17 CN CN201810783704.9A patent/CN110730280B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5923775A (en) * | 1996-04-04 | 1999-07-13 | Eastman Kodak Company | Apparatus and method for signal dependent noise estimation and reduction in digital images |
| EP1309185A1 (en) * | 2001-10-31 | 2003-05-07 | Sony International (Europe) GmbH | Image noise detection |
| JP2008131529A (en) * | 2006-11-24 | 2008-06-05 | Sony Corp | Imaging apparatus, and image processing apparatus, method and program |
| CN102801943A (en) * | 2011-05-24 | 2012-11-28 | 联发科技股份有限公司 | Methods for performing adaptive equalization and associated apparatus |
| CN103177423A (en) * | 2011-10-07 | 2013-06-26 | 伊姆普斯封闭式股份有限公司 | Method of noise reduction in digital x-ray frames series |
| CN102609914A (en) * | 2012-01-17 | 2012-07-25 | 天津大学 | Signal-correlated noise estimating method for image sensor |
| CN102968770A (en) * | 2012-11-30 | 2013-03-13 | 华为技术有限公司 | Method and device for eliminating noise |
| CN107993200A (en) * | 2017-11-02 | 2018-05-04 | 天津大学 | Picture noise level estimation method based on deep learning |
Non-Patent Citations (3)
| Title |
|---|
| Adaptive suppression of CCD signal-dependent noise in light space;H.Faraji et.al;《2004 IEEE International Conference on Acoustic，Speech，and Processing》;20040521;全文 * |
| Transformation of image-signal-dependent noise into image-signal-independent noise;Paul R. Prucbal et.al;《Optics Letters》;19810731;全文 * |
| 基于稀疏表示的CCD噪声估计及滤除;吴兆阳;《中国优秀硕士学位论文全文数据库 信息科技辑》;20140715;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110730280A (en) | 2020-01-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2013055492A1 (en) | Use of noise-optimized selection criteria to calculate scene white points | |
| CN104980652A (en) | Image processing apparatus and image processing method | |
| US9299011B2 (en) | Signal processing apparatus, signal processing method, output apparatus, output method, and program for learning and restoring signals with sparse coefficients | |
| CN112651903B (en) | Thermal infrared imager image preprocessing system and preprocessing method thereof | |
| KR20070078463A (en) | Apparatus and method for reducing noise from image sensor | |
| TWI820851B (en) | Compensation lookup table configuration method, display panel compensation method, compensation lookup table configuration device | |
| CN109697698B (en) | Low illuminance enhancement processing method, apparatus and computer readable storage medium | |
| US9761003B2 (en) | Stereo image depth map generation device and method | |
| US20220414896A1 (en) | Image processing method, electronic device and storage medium | |
| CN113538223A (en) | Noise image generation method, noise image generation device, electronic device, and storage medium | |
| CN104202448A (en) | System and method for solving shooting brightness unevenness of mobile terminal camera | |
| CN112788322B (en) | Adaptive white balance processing method, device, medium, and electronic apparatus | |
| US11823352B2 (en) | Processing video frames via convolutional neural network using previous frame statistics | |
| CN114445290A (en) | Hardware-oriented combined denoising and demosaicing method | |
| JP7024713B2 (en) | Image processing device and image processing method | |
| CN107113412A (en) | Display device, display calibration equipment, display correction system and display bearing calibration | |
| CN110730280B (en) | Noise equalization method and noise removal method | |
| US8744207B2 (en) | Image processing device, image processing method, image processing program and recording medium | |
| TWI689890B (en) | Noise equalization method and de-noise method | |
| JP2020181402A (en) | Image processing system, image processing method and program | |
| CN112950515B (en) | Image processing method and device, computer readable storage medium and electronic equipment | |
| CN106897980B (en) | Self-adaptive space domain noise reduction method based on local variance | |
| CN115334294A (en) | Video noise reduction method of local self-adaptive strength | |
| CN107124617B (en) | Method and system for generating random vector in motion estimation motion compensation | |
| CN114881886A (en) | Image processing method, image processing device, electronic equipment and storage medium |
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 |