CN106127712B - Image enhancement method and device - Google Patents
Image enhancement method and device Download PDFInfo
- Publication number
- CN106127712B CN106127712B CN201610510790.7A CN201610510790A CN106127712B CN 106127712 B CN106127712 B CN 106127712B CN 201610510790 A CN201610510790 A CN 201610510790A CN 106127712 B CN106127712 B CN 106127712B
- Authority
- CN
- China
- Prior art keywords
- layer
- image
- low
- frequency
- decomposition
- 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
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 160
- 230000009466 transformation Effects 0.000 claims abstract description 83
- 230000008569 process Effects 0.000 claims description 18
- 230000002708 enhancing effect Effects 0.000 claims description 7
- 230000010355 oscillation Effects 0.000 description 23
- 238000003745 diagnosis Methods 0.000 description 13
- 238000005070 sampling Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000005855 radiation Effects 0.000 description 2
- 238000002601 radiography Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10116—X-ray image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20016—Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20064—Wavelet transform [DWT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20172—Image enhancement details
- G06T2207/20192—Edge enhancement; Edge preservation
Abstract
An image enhancement method and device, the image enhancement method comprising: decomposing the image into L layers based on a first transformation; decomposing the image into L + N layers based on a second transformation, wherein N is more than or equal to 1; and reconstructing based on the low-frequency image of the L-th layer obtained by the second transformation decomposition and the high-frequency images of the first layer to the L-th layer obtained by the first transformation decomposition. The technical scheme of the invention enhances the contrast of the image to a great extent and improves the quality of the image.
Description
Technical Field
The present invention relates to the field of image processing technologies, and in particular, to an image enhancement method and apparatus.
Background
Digital Radiography (DR) equipment is an advanced medical equipment formed by combining computer Digital image processing technology and X-ray radiation technology. Digital X-ray photography equipment is widely used because of its small radiation dose, high image quality, high disease detection rate and high diagnostic accuracy.
For digital radiography equipment, the contrast of the image directly output by the detector is low, which is not beneficial for doctors to observe the focus and some fine details. In order to enable doctors to diagnose the illness condition of patients more conveniently and accurately, an image enhancement method is generally adopted in an image post-processing algorithm of the DR equipment to increase the contrast of an image, so that the details of the image are more prominent and clear. At present, an image enhancement method based on multi-scale is generally adopted to enhance an image acquired by a DR device, such as a wavelet transform, a laplacian pyramid image enhancement method, and the like. However, after the acquired image is processed by the current enhancement method, the quality of the enhanced image is not high, and the enhanced image does not meet the actual clinical requirements, so that the diagnosis of a doctor is inconvenient, and meanwhile, missed diagnosis or misdiagnosis can be caused.
Therefore, how to enhance the image to improve the image quality is one of the problems to be solved at present.
Disclosure of Invention
The invention aims to provide an image enhancement method and an image enhancement device to improve the contrast of an image and further improve the quality of the image.
In order to solve the above problem, the technical solution of the present invention provides an image enhancement method, including:
decomposing the image into L layers based on a first transformation;
decomposing the image into L + N layers based on a second transformation, wherein N is more than or equal to 1;
and reconstructing based on the low-frequency image of the L-th layer obtained by the second transformation decomposition and the high-frequency images of the first layer to the L-th layer obtained by the first transformation decomposition.
Optionally, reconstructing the low-frequency image of the L-th layer obtained based on the second transform decomposition and the high-frequency images of the first layer to the L-th layer obtained based on the first transform decomposition includes:
updating the low-frequency image of the L-th layer obtained by the first transformation decomposition by using the low-frequency image of the L-th layer obtained by the second transformation decomposition;
and reconstructing the high-frequency images from the first layer to the L-th layer obtained by the first transformation decomposition and the updated low-frequency image of the L-th layer.
Optionally, reconstructing the high-frequency image of the first layer to the L-th layer obtained by the first transform decomposition and the updated low-frequency image of the L-th layer includes:
carrying out bilinear interpolation or cubic interpolation on the updated low-frequency image of the L-i layer to obtain a first image of the L-i layer, and obtaining an updated low-frequency image of the L-i-1 layer based on the first image of the L-i layer and the high-frequency image of the L-i layer, wherein i is more than or equal to 0 and less than or equal to L-2;
repeating the above process until obtaining the updated low-frequency image of the first layer;
and reconstructing by using the high-frequency image of the first layer and the updated low-frequency image of the first layer.
Optionally, reconstructing the low-frequency image of the L-th layer obtained based on the second transform decomposition and the high-frequency images of the first layer to the L-th layer obtained based on the first transform decomposition includes:
respectively enhancing the high-frequency images from the first layer to the L-th layer obtained by the first transformation decomposition;
updating the low-frequency image of the L-th layer obtained by the first transformation decomposition by using the low-frequency image of the L-th layer obtained by the second transformation decomposition;
and reconstructing the enhanced high-frequency images of the first layer to the L-th layer and the updated low-frequency image of the L-th layer.
Optionally, the reconstructing with the enhanced high-frequency images of the first layer to the lth layer and the updated low-frequency image of the lth layer includes:
carrying out bilinear interpolation or cubic interpolation on the updated low-frequency image of the L-i layer to obtain a second image of the L-i layer, and obtaining an updated low-frequency image of the L-i-1 layer on the basis of the second image of the L-i layer and the enhanced high-frequency image of the L-i layer, wherein i is more than or equal to 0 and less than or equal to L-2;
repeating the above process until obtaining the updated low-frequency image of the first layer;
and reconstructing by using the enhanced high-frequency image of the first layer and the updated low-frequency image of the first layer.
Optionally, the low-frequency image of the L-th layer obtained by the second transform decomposition is obtained by reconstructing the low-frequency image of the L + N-th layer obtained by the second transform decomposition and the enhanced high-frequency images of the L + 1-th to L + N-th layers.
Optionally, the reconstructing the low-frequency image of the L + N th layer and the enhanced high-frequency images of the L +1 th to L + N th layers obtained by the second transform decomposition includes:
carrying out bilinear interpolation or cubic interpolation on the low-frequency image of the L + N-i layer to obtain a third image of the L + N-i layer, reconstructing the low-frequency image of the L + N-i-1 layer based on the third image of the L + N-i layer and the enhanced high-frequency image of the L + N-i layer, wherein i is more than or equal to 0 and less than or equal to N-1;
and repeating the process until the low-frequency image of the L-th layer is reconstructed.
Optionally, the first transform is a laplace transform, and the second transform is a wavelet transform.
In order to solve the above problem, the present invention further provides an image enhancement apparatus, including:
a first decomposition unit configured to decompose the image into L layers based on a first transform;
a second decomposition unit for decomposing the image into L + N layers based on a second transformation, N being equal to or greater than 1;
and the reconstruction unit is used for reconstructing based on the low-frequency image of the L-th layer obtained by the second transformation decomposition and the high-frequency images of the first layer to the L-th layer obtained by the first transformation decomposition.
Optionally, the first decomposition unit includes a laplace transform module, and the second decomposition unit includes a wavelet transform module.
Compared with the prior art, the technical scheme of the invention has the following advantages:
the image is decomposed into L layers through first transformation, the image is decomposed into L + N layers through second transformation, N is larger than or equal to 1 layer, and the low-frequency image of the L layer obtained through the second transformation decomposition and the high-frequency image from the first layer to the L layer obtained through the first transformation decomposition are reconstructed. For medical images, the quality of the enhanced images meets the actual clinical requirements, and the probability of missed diagnosis or misdiagnosis is reduced to a certain extent as the interior and the edge of the images are correspondingly enhanced.
Further, a second transform decomposition is performed to obtain a low frequency image of the L < th > layer by reconstructing the low frequency image of the L < th > layer and the enhanced high frequency images of the L < th > 1 < th > layer to the L < th > layer. By adopting the method, the low-frequency image of the L-th layer obtained by the second transformation decomposition is obtained, the contrast of the image obtained by final reconstruction can be further enhanced, and the quality of the image is further improved.
Furthermore, in the process of reconstructing the low-frequency image of the L-th layer obtained by the second transformation decomposition and the high-frequency images of the first to L-th layers obtained by the first transformation decomposition, the high-frequency images of the first to L-th layers obtained by the first transformation decomposition are enhanced, and finally the low-frequency images of the L-th layer updated in the first transformation and the high-frequency images of the enhanced first to L-th layers are reconstructed.
Furthermore, when reconstructing the low-frequency image of the L +1 th layer to the L + N th layer obtained by the second transformation decomposition, the interpolation mode of the low-frequency images of the L +1 th layer to the L + N th layer is changed, so that oscillation artifacts can be avoided, when reconstructing the low-frequency image of the first layer to the L th layer obtained by the updated first transformation decomposition and the high-frequency image of the first layer to the L th layer obtained by the first transformation decomposition, the interpolation mode of the low-frequency image of each updated layer is changed, so that the oscillation artifacts existing in the reconstructed image can be removed, and the oscillation artifacts which may exist in the second transformation and the first transformation decomposition processes are removed, so that the oscillation artifacts existing in the image are removed while the image contrast is improved, and the image quality is further improved. For the medical image, the finally obtained medical image is made to be more in line with the actual clinical requirement, and the occurrence of missed diagnosis or misdiagnosis is better reduced.
Drawings
FIG. 1 is a schematic flow chart of an image enhancement method according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of an image enhancement apparatus according to an embodiment of the present invention.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below. In the following description, specific details are set forth in order to provide a thorough understanding of the present invention. The invention can be implemented in a number of ways different from those described herein and similar generalizations can be made by those skilled in the art without departing from the spirit of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
As described in the prior art, when the image is enhanced by using the current image enhancement method, the obtained image quality is not ideal, for example, when the image is enhanced by using wavelet transform or laplace transform, the contrast of the enhanced image is not obvious, and for medical images, the enhanced image may not meet the actual clinical requirements, resulting in missed diagnosis or misdiagnosis. The inventor has found through analysis that when the wavelet transform is used for image enhancement, fine details of an image can be well enhanced, but the enhancement of the edge contour of the image is not obvious, and in addition, when the wavelet transform is used for enhancing the image, oscillation artifacts can also appear in the image, so that the quality of the obtained image is not high when the wavelet transform is used for enhancing the image. When the image is enhanced by the laplace transform, the edge contour of the image is obviously enhanced, but the fine detail part of the image is not obviously enhanced, and when the image is enhanced by the laplace transform, oscillation artifacts exist, and the quality of the image is not high. Therefore, the inventor considers whether the wavelet transform and the laplacian transform can be combined, so that the finally obtained image can be obviously enhanced on fine details and edge contours, namely, the contrast of the image is enhanced; and in addition, oscillation artifacts generated in the image enhancement process by the wavelet transform and the Laplace transform are correspondingly removed.
Referring to fig. 1, fig. 1 is a schematic flow chart of an image enhancement method according to an embodiment of the present invention, and as shown in fig. 1, the image enhancement method includes:
s11: decomposing the image into L layers based on a first transformation;
s12: decomposing the image into L + N layers based on a second transformation, wherein N is more than or equal to 1;
s13: and reconstructing based on the low-frequency image of the L-th layer obtained by the second transformation decomposition and the high-frequency images of the first layer to the L-th layer obtained by the first transformation decomposition.
In the present embodiment, the first transformation may be performed after the image is decomposed by the first transformation, and the detail information of the image is mainly concentrated in the high-frequency image and the low-frequency image of the first layers obtained by the first transformation decomposition. The image may be enhanced based on the above enhancement method by using a first transform or a second transform as long as the transform satisfies the above characteristics. In the present embodiment, the image reconstruction is performed based on the high-frequency image of the first layer to the L-th layer obtained by the first transformation and the low-frequency image of the L-th layer obtained by the second transformation, and both the fine details in the image and the edge of the image are enhanced.
The image enhancement method according to the embodiment of the present invention will be described in detail below with reference to examples of the first transform and the second transform, i.e., laplace transform. However, the first transformation and the second transformation are not limited in the technical solution of the present invention, and any transformation that satisfies the above-described characteristics may be used as the first transformation or the second transformation.
Example one
Execution of S11: and carrying out pyramid decomposition on the image through Laplace transform to obtain a high-frequency image and a low-frequency image of each layer of the image. For example, if the image is decomposed into three layers by using the laplace transform, the image is decomposed into a low-frequency image of a first layer and a high-frequency image of the first layer, then the low-frequency image of the first layer is decomposed into a low-frequency image of a second layer and a high-frequency image of the second layer, and then the low-frequency image of the second layer is decomposed into a low-frequency image of a third layer and a high-frequency image of the third layer. In this embodiment, the image is decomposed into several layers by using laplacian transform, which may be determined according to the actual size of the image and the actual requirements during the processing, and generally, when the image is pyramid-decomposed by using laplacian transform, the scales of the previous layers are smaller, the details of the included image are more, and in addition, considering the timeliness during the actual processing, the image may be decomposed into three layers by using laplacian transform in this embodiment, that is, L is 3.
Execution of S12: and decomposing the image by adopting wavelet transform, wherein the number of layers for decomposing the image by adopting wavelet transform in the step is larger than that for decomposing the image by adopting Laplace transform. Still taking the case of decomposing the image into three layers by using the laplace transform, the image may be decomposed into four layers by using the wavelet transform, that is, the image is decomposed into the low-frequency image of the first layer and the high-frequency image of the first layer by using the wavelet transform, then the low-frequency image of the first layer is decomposed into the low-frequency image of the second layer and the high-frequency image of the second layer, the low-frequency image of the second layer is decomposed into the low-frequency image of the third layer and the high-frequency image of the third layer, and finally the low-frequency image of the third layer is decomposed into the low-frequency image of the fourth layer and the high-frequency image of the fourth layer. The number of layers for decomposing the image by adopting the wavelet transform can be determined according to actual requirements on the premise of ensuring that the number of layers is larger than that for decomposing the image by adopting the Laplace transform.
S13 is performed, and the low-frequency image of the L-th layer obtained by the wavelet transform decomposition and the high-frequency images of the first to L-th layers obtained by the laplace transform decomposition are reconstructed to obtain an enhanced image. The present embodiment performs reconstruction to obtain an enhanced image as follows.
As can be seen from the above description, by performing S11 and S12 to decompose the image into L layers (first to L-th layers) by laplace transform and L + N layers (first to L + N layers) by wavelet transform, in the present embodiment, the low-frequency image of the L-th layer obtained by laplace transform decomposition is updated with the low-frequency image of the L-th layer obtained by wavelet transform decomposition, that is, the low-frequency image of the L-th layer obtained by laplace transform decomposition is replaced with the low-frequency image of the L-th layer obtained by wavelet transform decomposition, and after the low-frequency image of the L-th layer obtained by laplace transform decomposition is updated, the low-frequency images of the L-1-th layer to the first layer before the low-frequency image are sequentially updated according to the updated low-frequency image of the L-th layer, that is to update the low-frequency image of the L-1-th layer first, that is to add the updated low-frequency image of the L-th layer to obtain the updated L-1-layer low-frequency image (ii) a Updating the low-frequency image of the L-2 layer, specifically adding the high-frequency image of the L-1 layer with the updated low-frequency image of the L-1 layer to obtain an updated low-frequency image of the L-2 layer; and the like until the low-frequency image of the first layer is updated. And then reconstructing an enhanced image by using the high-frequency image of the first layer obtained by the Laplace transform decomposition and the low-frequency image of the first layer obtained by the updated Laplace transform decomposition.
In this embodiment, the low-frequency image of the L-th layer obtained by wavelet transform decomposition may be obtained by performing linear or nonlinear enhancement on the high-frequency images of the L + 1-th to L + N-th layers obtained by wavelet transform decomposition to obtain high-frequency images of the L + 1-th to L + N-th layers, and then reconstructing the high-frequency images with the low-frequency image of the L + N-th layer obtained by wavelet transform decomposition. The low-frequency image of the L-th layer obtained by wavelet transform decomposition is obtained in the mode, the contrast of the image obtained by final reconstruction can be further enhanced, and the quality of the image is improved.
In other embodiments, the low-frequency image of the L-th layer obtained by wavelet transform decomposition may also be directly reconstructed from the low-frequency image of the L + N-th layer and the high-frequency images of the L + 1-L + N-th layers.
The process of reconstructing an image based on laplace and wavelet transform to obtain an enhanced image in the present embodiment is briefly described below by taking an example in which the number of layers for decomposing an image using laplace transform is three, and the number of layers for decomposing an image using wavelet transform is five.
Firstly, the image is decomposed into three layers by adopting Laplace transform, and the image is decomposed into five layers by adopting wavelet transform. And then updating the low-frequency image of the third layer obtained by the decomposition of the Laplace transform with the low-frequency image of the third layer obtained by the decomposition of the wavelet transform.
The low-frequency image of the third layer obtained by wavelet transform decomposition is obtained by reconstructing the low-frequency image of the fifth layer obtained by wavelet transform decomposition and the high-frequency images of the fourth layer and the fifth layer obtained by respectively carrying out nonlinear enhancement on the high-frequency images of the fourth layer and the fifth layer.
After the low-frequency image of the third layer obtained by wavelet transform decomposition is obtained, the low-frequency image of the third layer obtained by the wavelet transform decomposition is used as a low-frequency image of the third layer obtained by Laplace transform decomposition, and the updated low-frequency images of the first layer and the second layer are determined by taking the low-frequency image of the third layer obtained by the Laplace transform decomposition as a reference; that is, the low-frequency images of the first layer to the second layer obtained by the laplace transform decomposition are sequentially recalculated based on the updated low-frequency image of the third layer and the high-frequency images of the first layer to the third layer obtained by the laplace transform decomposition, and the enhanced image is reconstructed by using the updated low-frequency image of the first layer and the high-frequency image of the first layer obtained by the laplace transform decomposition.
In the embodiment, the information of the front L layer obtained by the Laplace transform decomposition and the information of the L + N layer obtained by the wavelet transform decomposition are effectively combined, so that the fine details and the edge contour in the image are correspondingly enhanced, and the image quality is improved. For medical images, the quality of the enhanced images meets the actual clinical requirements, and the probability of missed diagnosis or misdiagnosis is reduced.
Example two
In this embodiment, the image is decomposed into L layers (first to L-th layers) by laplace transform, and the image is decomposed into L + N layers (first to L + N-th layers) by wavelet transform, which is similar to that in the first embodiment and will not be described herein again. The difference between this embodiment and the first embodiment is that, when reconstructing the high-frequency images of the first to L-th layers obtained by the laplace transform decomposition and the low-frequency images of the L-th layer obtained by the wavelet transform decomposition, in order to further enhance the contrast of the reconstructed images, in this embodiment, the high-frequency images of the first to L-th layers obtained by the laplace transform decomposition are respectively enhanced, and specifically, the high-frequency images of the first to L-th layers may be respectively enhanced linearly or nonlinearly. The updating of the low-frequency image of the L-th layer obtained by the laplace transform decomposition is similar to that in the first embodiment, that is, the low-frequency image of the L-th layer obtained by the laplace transform decomposition is updated by the low-frequency image of the L-th layer obtained by the wavelet transform decomposition, and the low-frequency images of the first layers of the L-1-th layer, the L-2-th layer and the L-3-th layer … … are sequentially updated based on the updated low-frequency image of the L-th layer. The low-frequency image of the L-th layer obtained by wavelet transform decomposition is also similar to that obtained in the first embodiment, and is not described herein again.
And finally reconstructing the high-frequency images of the first layer to the L layer obtained by enhancing the high-frequency images of the first layer to the L layer obtained by the Laplace transform decomposition and the updated low-frequency image of the L layer to obtain an enhanced image. In this embodiment, the low-frequency image of the L-th layer obtained by wavelet transform decomposition may be obtained by reconstructing a low-frequency image of the L + N-th layer obtained by wavelet transform decomposition and a high-frequency image of the L + 1-L + N-th layers obtained by enhancing a high-frequency image of the L + 1-L + N-th layers obtained by wavelet transform decomposition, and updating the low-frequency image of the L-th layer obtained by laplace transform decomposition with the low-frequency image of the L-th layer obtained by reconstruction, and finally reconstructing the low-frequency image of the L-th layer updated by laplace transform and the high-frequency image of the first-L-th layers obtained by enhanced laplace transform decomposition, so as to greatly enhance the fine details and edge contours in the image, improve the quality of the reconstructed image, and for the medical image, the quality of the reconstructed image meets the actual clinical requirement, which is helpful for the diagnosis of doctors and reduces the probability of missed diagnosis or misdiagnosis.
EXAMPLE III
In this embodiment, an image is decomposed into L layers (first to L-th layers) by laplace transform, the image is decomposed into L + N layers (first to L + N layers) by wavelet transform, a low-frequency image of the L layer obtained by laplace transform decomposition is updated with a low-frequency image of the L layer obtained by wavelet transform decomposition, and corresponding updating is performed on low-frequency images of the L-1-th to first layers in sequence with the updated low-frequency image of the L layer, which is similar to that in embodiment one, and details are not repeated here. The difference between this embodiment and the first and second embodiments is that in this embodiment, it is considered that oscillation artifacts exist when images of each layer obtained by wavelet transform and laplace transform decomposition are reconstructed, and the oscillation artifacts have an influence on the quality of the images, so in this embodiment, oscillation artifacts generated during reconstruction by laplace transform and wavelet transform are removed accordingly, and the following description is made accordingly.
In this embodiment, the low-frequency image of the L-th layer obtained by wavelet transform decomposition is obtained by reconstructing the low-frequency image of the L + N-th layer obtained by wavelet transform decomposition and the enhanced high-frequency images of the L + 1-th to L + N-th layers; the skilled person knows that when an image is decomposed by using wavelet transform or laplace transform, it is necessary to downsample the image, so that it is necessary to upsample the image in the process of reconstruction, and at present, a method of interpolating 0 in a low-frequency image is generally used to upsample the image, and the inventors have studied and found that upsampling the low-frequency image by using the method of interpolating 0 can cause oscillation artifacts. Therefore, in order to eliminate oscillation artifacts generated by wavelet transformation, in the process of reconstructing the low-frequency image of the L + N layer, the low-frequency image of the L + N layer obtained by wavelet transformation decomposition is subjected to up-sampling in a bilinear interpolation mode to obtain a third image of the L + N layer, and the low-frequency image of the L + N layer obtained by wavelet transformation decomposition is reconstructed on the basis of the third image of the L + N layer and the high-frequency image of the L + N layer obtained by wavelet transformation decomposition after the high-frequency image of the L + N layer is enhanced; then, the reconstructed low-frequency image of the L + N-1 layer is continuously subjected to up-sampling in a bilinear interpolation mode to obtain a third image of the L + N-2 layer, and a high-frequency image of the L + N-2 layer obtained by wavelet transformation decomposition is reconstructed on the basis of the third image of the L + N-2 layer and the high-frequency image of the L + N-2 layer obtained by wavelet transformation decomposition to reconstruct a low-frequency image of the L + N-3 layer obtained by wavelet transformation decomposition, and so on until the low-frequency image of the L layer obtained by wavelet transformation decomposition is reconstructed. The upsampling by means of bilinear interpolation is prior art in the art and will not be described herein. In other embodiments, the up-sampling of the low-frequency images of the L +1 th layer to the L + N th layer may also be implemented by performing three times of interpolation on the low-frequency images of the L +1 th layer to the L + N th layer, respectively.
When the low-frequency image of the L-th layer obtained by wavelet transform decomposition is reconstructed, the interpolation mode of the low-frequency images from the L + 1-th layer to the L + N-th layer is changed, so that oscillation artifacts can be avoided in the low-frequency image of the L-th layer, and further oscillation artifacts can be avoided in the finally reconstructed image due to the oscillation artifacts when the low-frequency image of the L-th layer obtained by Laplace transform decomposition is updated with the low-frequency image of the L-th layer.
After removing the ringing artifact in the low-frequency image of the L-th layer obtained by the wavelet transform decomposition, the low-frequency image of the L-th layer obtained by the laplace transform decomposition and the low-frequency images of the preceding L-1 th to first layers are updated with the low-frequency image of the L-th layer obtained by the wavelet transform decomposition in the same manner. In this embodiment, since the oscillation artifact may exist in the laplace transform, when reconstructing the low-frequency image of the L-th layer obtained by the updated laplace transform decomposition and the high-frequency images of the first to L-th layers obtained by enhancing the high-frequency images of the first to L-th layers obtained by the laplace transform decomposition, it is still necessary to change the method of interpolating the low-frequency images of the L-th to first layers to reduce the occurrence of the oscillation artifact. In the same way, namely in the reconstruction process, the updated low-frequency image of the L-th layer is up-sampled in a bilinear interpolation mode to obtain a second image of the L-th layer, the updated low-frequency image of the L-1-th layer is obtained based on the second image of the L-th layer and the enhanced high-frequency image of the L-th layer, and then continuously performing up-sampling on the updated low-frequency image of the L-1 layer in a bilinear interpolation mode to obtain a second image of the L-1 layer, obtaining an updated low-frequency image of the L-2 layer based on the second image of the L-1 layer and the enhanced high-frequency image of the L-1 layer, and so on until obtaining an updated low-frequency image of the first layer, and finally reconstructing the high-frequency image of the first layer obtained by decomposition of the enhanced Laplace transform and the low-frequency image of the first layer obtained by decomposition of the updated Laplace transform. The upsampling by means of bilinear interpolation is prior art in the art and will not be described herein. In other embodiments, the up-sampling of the updated low-frequency images of the first layer to the L-th layer may also be implemented by performing three times of interpolation on the updated low-frequency images of the first layer to the L-th layer, respectively.
When reconstructing the updated L-layer low-frequency image obtained by Laplace transform decomposition and the first-layer to L-layer high-frequency images obtained by Laplace transform decomposition, changing the interpolation mode of the updated low-frequency image of each layer can remove oscillation artifacts existing in the reconstructed image, further improve the quality of the image, enable the finally obtained medical image to better meet the actual clinical requirement for the medical image, and reduce the occurrence of missed diagnosis or misdiagnosis.
The process of reconstructing an image based on laplace and wavelet transform to obtain an enhanced image in the present embodiment is briefly described below by taking an example in which the number of layers for decomposing an image using laplace transform is three, and the number of layers for decomposing an image using wavelet transform is five.
Firstly, the image is decomposed into three layers by adopting Laplace transform, and the image is decomposed into five layers by adopting wavelet transform. And then updating the low-frequency image of the third layer obtained by the decomposition of the Laplace transform with the low-frequency image of the third layer obtained by the decomposition of the wavelet transform.
The low-frequency image of the third layer obtained by wavelet transformation decomposition is obtained by performing bilinear interpolation on the low-frequency image of the fifth layer obtained by wavelet transformation decomposition, so as to reconstruct the low-frequency image of the fourth layer by performing bilinear interpolation on the low-frequency image of the fifth layer and the high-frequency image of the fifth layer subjected to nonlinear enhancement, and then reconstructing the low-frequency image of the third layer obtained by wavelet transformation decomposition by using the low-frequency image of the fourth layer subjected to bilinear interpolation and the high-frequency image of the fourth layer subjected to nonlinear enhancement.
After the low-frequency image of the third layer obtained by wavelet transform decomposition is obtained, the low-frequency image is used as the low-frequency image of the third layer obtained by Laplace transform decomposition, and updating the low-frequency images of the first layer and the second layer obtained by the Laplace transform decomposition by taking the low-frequency images as a reference, specifically, firstly carrying out bilinear interpolation on the low-frequency image of the third layer obtained by the updated Laplace transform decomposition, adding the high-frequency image of the third layer obtained by nonlinear enhanced Laplace transform decomposition to the low-frequency image of the third layer subjected to bilinear interpolation to obtain a low-frequency image of the second layer, then adding the low-frequency image of the second layer subjected to bilinear interpolation to the high-frequency image of the second layer obtained by nonlinear enhanced Laplace transform decomposition to obtain a low-frequency image of the first layer, the enhanced image is reconstructed by adding the low frequency image of the first layer to the high frequency image of the first layer resulting from the non-linearly enhanced laplacian transform decomposition.
In the present embodiment, the information of the first L layer obtained by the laplace transform decomposition and the information of the L + N th layer obtained by the wavelet transform decomposition are effectively combined, and when the low-frequency image of the L-th layer obtained by the wavelet transform decomposition is obtained, the low-frequency image and the enhanced high-frequency image of the subsequent layers are reconstructed, and then the image reconstruction is realized by the updated low-frequency image information of the L-th layer obtained by the laplace transform decomposition and the enhanced high-frequency images of the first to L-th layers, so that the fine details and the edge contour in the image are enhanced to a large extent. In addition, in the process of reconstructing the low-frequency image of the L layer obtained by wavelet transform decomposition, the interpolation mode is changed, and in the process of reconstructing the low-frequency image of the L layer obtained by updated Laplace transform decomposition and the high-frequency images of the first layer to the L layer obtained by enhanced Laplace transform decomposition, the interpolation mode of the low-frequency image of the first layer to the L layer obtained by updated Laplace transform decomposition is changed, so that the finally obtained image has higher contrast and does not have oscillation artifacts, the quality of the reconstructed image is improved to a great extent, for the medical image, the quality of the reconstructed image is more in line with the actual clinical requirement, the diagnosis of a doctor is facilitated, and the possibility of missed diagnosis or misdiagnosis is reduced.
Corresponding to the image enhancement method, an image enhancement apparatus according to an embodiment of the present invention is further provided, referring to fig. 2, and fig. 2 is a schematic structural diagram of the image enhancement apparatus according to the embodiment of the present invention, as shown in fig. 2, the image enhancement apparatus includes:
a first decomposition unit 10 for decomposing the image into L layers based on a first transformation;
a second decomposition unit 11, configured to decompose the image into L + N layers based on a second transformation, where N is greater than or equal to 1;
a reconstruction unit 12, configured to perform reconstruction based on the low-frequency image of the L-th layer obtained by the second transform decomposition and the high-frequency images of the first to L-th layers obtained by the first transform decomposition.
In an embodiment, the first decomposition unit comprises a laplacian transform module and the second decomposition unit comprises a wavelet transform module.
The specific implementation of the image enhancement apparatus may refer to the implementation of the image enhancement method, and is not described herein again.
In summary, the image enhancement method provided by the embodiment of the present invention at least has the following beneficial effects:
the image is decomposed into L layers through first transformation, the image is decomposed into L + N layers through second transformation, N is larger than or equal to 1 layer, and the low-frequency image of the L layer obtained through the second transformation decomposition and the high-frequency image from the first layer to the L layer obtained through the first transformation decomposition are reconstructed. For medical images, the quality of the enhanced images meets the actual clinical requirements, and the probability of missed diagnosis or misdiagnosis is reduced to a certain extent as the interior and the edge of the images are correspondingly enhanced.
Further, a second transform decomposition is performed to obtain a low frequency image of the L < th > layer by reconstructing the low frequency image of the L < th > layer and the enhanced high frequency images of the L < th > 1 < th > layer to the L < th > layer. By adopting the method, the low-frequency image of the L-th layer obtained by the second transformation decomposition is obtained, the contrast of the image obtained by final reconstruction can be further enhanced, and the quality of the image is further improved.
Furthermore, in the process of reconstructing the low-frequency image of the L-th layer obtained by the second transformation decomposition and the high-frequency images of the first to L-th layers obtained by the first transformation decomposition, the high-frequency images of the first to L-th layers obtained by the first transformation decomposition are enhanced, and finally the low-frequency images of the L-th layer updated in the first transformation and the high-frequency images of the enhanced first to L-th layers are reconstructed.
Furthermore, when reconstructing the low-frequency image of the L +1 th layer to the L + N th layer obtained by the second transformation decomposition, the interpolation mode of the low-frequency images of the L +1 th layer to the L + N th layer is changed, so that oscillation artifacts can be avoided, when reconstructing the low-frequency image of the first layer to the L th layer obtained by the updated first transformation decomposition and the high-frequency image of the first layer to the L th layer obtained by the first transformation decomposition, the interpolation mode of the low-frequency image of each updated layer is changed, so that the oscillation artifacts existing in the reconstructed image can be removed, and the oscillation artifacts which may exist in the second transformation and the first transformation decomposition processes are removed, so that the oscillation artifacts existing in the image are removed while the image contrast is improved, and the image quality is further improved. For the medical image, the finally obtained medical image is made to be more in line with the actual clinical requirement, and the occurrence of missed diagnosis or misdiagnosis is better reduced.
Although the present invention has been described with reference to the preferred embodiments, it is not intended to limit the present invention, and those skilled in the art can make variations and modifications of the present invention without departing from the spirit and scope of the present invention by using the methods and technical contents disclosed above.
Claims (8)
1. An image enhancement method, comprising:
decomposing the image into L layers based on a first transformation;
decomposing the image into L + N layers based on a second transformation, wherein N is more than or equal to 1;
reconstructing based on the low-frequency image of the L-th layer obtained by the second transformation decomposition and the high-frequency images of the first layer to the L-th layer obtained by the first transformation decomposition;
wherein the first transform is a laplace transform and the second transform is a wavelet transform.
2. The image enhancement method according to claim 1, wherein the reconstructing based on the low-frequency image of the L-th layer obtained by the second transform decomposition and the high-frequency images of the first to L-th layers obtained by the first transform decomposition comprises:
updating the low-frequency image of the L-th layer obtained by the first transformation decomposition by using the low-frequency image of the L-th layer obtained by the second transformation decomposition;
and reconstructing the high-frequency images from the first layer to the L-th layer obtained by the first transformation decomposition and the updated low-frequency image of the L-th layer.
3. The image enhancement method according to claim 2, wherein the reconstructing the first layer to lth layer high-frequency image and the updated lth layer low-frequency image obtained by the first transform decomposition comprises:
carrying out bilinear interpolation or cubic interpolation on the updated low-frequency image of the L-i layer to obtain a first image of the L-i layer, and obtaining an updated low-frequency image of the L-i-1 layer based on the first image of the L-i layer and the high-frequency image of the L-i layer, wherein i is more than or equal to 0 and less than or equal to L-2;
repeating the above process until obtaining the updated low-frequency image of the first layer;
and reconstructing by using the high-frequency image of the first layer and the updated low-frequency image of the first layer.
4. The image enhancement method according to claim 1, wherein the reconstructing based on the low-frequency image of the L-th layer obtained by the second transform decomposition and the high-frequency images of the first to L-th layers obtained by the first transform decomposition comprises:
respectively enhancing the high-frequency images from the first layer to the L-th layer obtained by the first transformation decomposition;
updating the low-frequency image of the L-th layer obtained by the first transformation decomposition by using the low-frequency image of the L-th layer obtained by the second transformation decomposition;
and reconstructing the enhanced high-frequency images of the first layer to the L-th layer and the updated low-frequency image of the L-th layer.
5. The image enhancement method according to claim 4, wherein the reconstructing with the enhanced high-frequency images of the first layer through the Lth layer and the updated low-frequency image of the Lth layer comprises:
carrying out bilinear interpolation or cubic interpolation on the updated low-frequency image of the L-i layer to obtain a second image of the L-i layer, and obtaining an updated low-frequency image of the L-i-1 layer on the basis of the second image of the L-i layer and the enhanced high-frequency image of the L-i layer, wherein i is more than or equal to 0 and less than or equal to L-2;
repeating the above process until obtaining the updated low-frequency image of the first layer;
and reconstructing by using the enhanced high-frequency image of the first layer and the updated low-frequency image of the first layer.
6. The image enhancement method according to any one of claims 2 to 5, wherein the low-frequency image of the L-th layer obtained by the second transform decomposition is obtained by reconstructing the low-frequency image of the L + N-th layer obtained by the second transform decomposition and the enhanced high-frequency images of the L + 1-th to L + N-th layers.
7. The image enhancement method of claim 6, wherein the reconstructing the low-frequency image of the L + N th layer and the enhanced high-frequency images of the L +1 th to L + N th layers, which are decomposed by the second transform, comprises:
carrying out bilinear interpolation or cubic interpolation on the low-frequency image of the L + N-i layer to obtain a third image of the L + N-i layer, reconstructing the low-frequency image of the L + N-i-1 layer based on the third image of the L + N-i layer and the enhanced high-frequency image of the L + N-i layer, wherein i is more than or equal to 0 and less than or equal to N-1;
and repeating the process until the low-frequency image of the L-th layer is reconstructed.
8. An image enhancement apparatus, comprising:
a first decomposition unit configured to decompose the image into L layers based on a first transform;
a second decomposition unit for decomposing the image into L + N layers based on a second transformation, N being equal to or greater than 1;
a reconstruction unit, configured to reconstruct a low-frequency image of an L-th layer obtained by the second transform decomposition and a high-frequency image of first to L-th layers obtained by the first transform decomposition;
wherein the first decomposition unit comprises a laplace transform module and the second decomposition unit comprises a wavelet transform module.
Priority Applications (13)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610510790.7A CN106127712B (en) | 2016-07-01 | 2016-07-01 | Image enhancement method and device |
| BR112018013602-6A BR112018013602A2 (en) | 2015-12-31 | 2016-12-29 | image processing methods and systems |
| CN201680083009.0A CN108780571B (en) | 2015-12-31 | 2016-12-29 | Image processing method and system |
| CN202210868915.9A CN115049563A (en) | 2015-12-31 | 2016-12-29 | Image processing method and system |
| RU2019134059A RU2797310C2 (en) | 2015-12-31 | 2016-12-29 | Methods and systems for image processing |
| GB1710525.5A GB2548767B (en) | 2015-12-31 | 2016-12-29 | Methods and systems for image processing |
| RU2018127759A RU2705014C1 (en) | 2015-12-31 | 2016-12-29 | Methods and systems for image processing |
| EP16881262.6A EP3398159B1 (en) | 2015-12-31 | 2016-12-29 | Methods and systems for image processing |
| PCT/CN2016/113079 WO2017114473A1 (en) | 2015-12-31 | 2016-12-29 | Methods and systems for image processing |
| EP21174529.4A EP3920133A1 (en) | 2015-12-31 | 2016-12-29 | Methods and systems for image processing |
| US15/638,327 US10290108B2 (en) | 2015-12-31 | 2017-06-29 | Methods and systems for image processing |
| US16/410,119 US11049254B2 (en) | 2015-12-31 | 2019-05-13 | Methods and systems for image processing |
| US17/361,315 US11880978B2 (en) | 2015-12-31 | 2021-06-28 | Methods and systems for image processing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610510790.7A CN106127712B (en) | 2016-07-01 | 2016-07-01 | Image enhancement method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106127712A CN106127712A (en) | 2016-11-16 |
| CN106127712B true CN106127712B (en) | 2020-03-31 |
Family
ID=57467810
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610510790.7A Active CN106127712B (en) | 2015-12-31 | 2016-07-01 | Image enhancement method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106127712B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115049563A (en) * | 2015-12-31 | 2022-09-13 | 上海联影医疗科技股份有限公司 | Image processing method and system |
| CN108986074B (en) * | 2018-06-12 | 2022-05-17 | 中国矿业大学 | Coal mine underground image processing method based on blocking sparse algorithm |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101030298A (en) * | 2007-03-29 | 2007-09-05 | 杭州电子科技大学 | Method for enhancing medical image with multi-scale self-adaptive contrast change |
| CN101109718A (en) * | 2006-11-14 | 2008-01-23 | 北京国药恒瑞美联信息技术有限公司 | Virtual grid imaging method and system used for eliminating influence of scattered radiation |
| CN101201937A (en) * | 2007-09-18 | 2008-06-18 | 上海医疗器械厂有限公司 | Digital image enhancement method and device based on wavelet restruction and decompose |
| CN101212560A (en) * | 2007-12-21 | 2008-07-02 | 上海广电集成电路有限公司 | Method for improving video image sharpness point by point |
| CN101609549A (en) * | 2009-07-24 | 2009-12-23 | 河海大学常州校区 | The multi-scale geometric analysis super-resolution processing method of video blurred image |
| CN101739671A (en) * | 2009-12-30 | 2010-06-16 | 中兴通讯股份有限公司 | Method and system for reducing noise of noisy images |
| CN102324021A (en) * | 2011-09-05 | 2012-01-18 | 电子科技大学 | Infrared dim-small target detection method based on shear wave conversion |
| CN102428478A (en) * | 2009-04-17 | 2012-04-25 | 里弗兰医疗集团公司 | Multi-scale image normalization and enhancement |
| CN103049898A (en) * | 2013-01-27 | 2013-04-17 | 西安电子科技大学 | Method for fusing multispectral and full-color images with light cloud |
| CN105931201A (en) * | 2016-04-20 | 2016-09-07 | 北京航空航天大学 | Image subjective visual effect enhancing method based on wavelet transformation |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9598011B2 (en) * | 2014-01-09 | 2017-03-21 | Northrop Grumman Systems Corporation | Artificial vision system |
-
2016
- 2016-07-01 CN CN201610510790.7A patent/CN106127712B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101109718A (en) * | 2006-11-14 | 2008-01-23 | 北京国药恒瑞美联信息技术有限公司 | Virtual grid imaging method and system used for eliminating influence of scattered radiation |
| CN101030298A (en) * | 2007-03-29 | 2007-09-05 | 杭州电子科技大学 | Method for enhancing medical image with multi-scale self-adaptive contrast change |
| CN101201937A (en) * | 2007-09-18 | 2008-06-18 | 上海医疗器械厂有限公司 | Digital image enhancement method and device based on wavelet restruction and decompose |
| CN101212560A (en) * | 2007-12-21 | 2008-07-02 | 上海广电集成电路有限公司 | Method for improving video image sharpness point by point |
| CN102428478A (en) * | 2009-04-17 | 2012-04-25 | 里弗兰医疗集团公司 | Multi-scale image normalization and enhancement |
| CN101609549A (en) * | 2009-07-24 | 2009-12-23 | 河海大学常州校区 | The multi-scale geometric analysis super-resolution processing method of video blurred image |
| CN101739671A (en) * | 2009-12-30 | 2010-06-16 | 中兴通讯股份有限公司 | Method and system for reducing noise of noisy images |
| CN102324021A (en) * | 2011-09-05 | 2012-01-18 | 电子科技大学 | Infrared dim-small target detection method based on shear wave conversion |
| CN103049898A (en) * | 2013-01-27 | 2013-04-17 | 西安电子科技大学 | Method for fusing multispectral and full-color images with light cloud |
| CN105931201A (en) * | 2016-04-20 | 2016-09-07 | 北京航空航天大学 | Image subjective visual effect enhancing method based on wavelet transformation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106127712A (en) | 2016-11-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Eo et al. | KIKI‐net: cross‐domain convolutional neural networks for reconstructing undersampled magnetic resonance images | |
| CN110060313B (en) | Image artifact correction method and system | |
| Park et al. | Computed tomography super-resolution using deep convolutional neural network | |
| US7929746B2 (en) | System and method for processing imaging data | |
| Yuan et al. | Half2Half: deep neural network based CT image denoising without independent reference data | |
| US20100092102A1 (en) | Method and apparatus for image processing | |
| US10867375B2 (en) | Forecasting images for image processing | |
| CN113920212B (en) | Magnetic resonance reconstruction model training method, computer device and storage medium | |
| Li et al. | A modified generative adversarial network using spatial and channel-wise attention for CS-MRI reconstruction | |
| CN106127712B (en) | Image enhancement method and device | |
| Hou et al. | CT image quality enhancement via a dual-channel neural network with jointing denoising and super-resolution | |
| Li et al. | Learning non-local perfusion textures for high-quality computed tomography perfusion imaging | |
| Sahu et al. | An application of deep dual convolutional neural network for enhanced medical image denoising | |
| CN112330565A (en) | Image denoising method in low-dose CT projection domain based on improved U-net | |
| CN111860836A (en) | Self-supervision learning method and application | |
| CN113592968B (en) | Method and device for reducing metal artifacts in tomographic images | |
| CN111462004B (en) | Image enhancement method and device, computer equipment and storage medium | |
| CN113053496B (en) | Deep learning method for low-dose estimation of medical image | |
| Jin et al. | Low-dose CT image restoration based on noise prior regression network | |
| WO2009013696A2 (en) | Framework system and method for low-frequency preservation in multiresolution nonlinear adaptive filtering | |
| Kazmi et al. | Medical image denoising based on adaptive thresholding in contourlet domain | |
| WO2022193276A1 (en) | Deep learning method for low dose estimation of medical image | |
| Liang et al. | A model-based deep learning reconstruction for X-Ray CT | |
| CN117274080B (en) | Low-dose CT chord graph recovery method and related device | |
| CN111815766B (en) | Processing method and system for reconstructing three-dimensional model of blood vessel based on 2D-DSA image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 201807 Shanghai city Jiading District Industrial Zone Jiading Road No. 2258 Patentee after: Shanghai Lianying Medical Technology Co., Ltd Address before: 201807 Shanghai city Jiading District Industrial Zone Jiading Road No. 2258 Patentee before: SHANGHAI UNITED IMAGING HEALTHCARE Co.,Ltd. |