CN1819621A - Medical image enhancing processing method - Google Patents
Medical image enhancing processing method Download PDFInfo
- Publication number
- CN1819621A CN1819621A CN 200610049357 CN200610049357A CN1819621A CN 1819621 A CN1819621 A CN 1819621A CN 200610049357 CN200610049357 CN 200610049357 CN 200610049357 A CN200610049357 A CN 200610049357A CN 1819621 A CN1819621 A CN 1819621A
- Authority
- CN
- China
- Prior art keywords
- image
- low pass
- original image
- high fdrequency
- processing method
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
A medicine image enhancement processing method, filtering by the low pass filter, convolute the original image, smooth the image to get the low pass smooth image, distill the high frequency component by decrease the low pass smooth image from the original image, furl to the original image, the character are: convolute the shrunken image by the fixed convoluted core, distill the high frequency component on the different space frequency strip, enlarge the high frequency component to the image with the original size, and furl to the original image. This processing method is that enhances the images on the different space frequency strip simultaneously in a short time, strengthen the organic edge and the detail information of the image effectively, satisfy the clinical need.
Description
Technical field
The present invention relates to a kind of image enhancement processing method, particularly a kind of medical image is carried out enhancement process, to highlight the method for organization edge and detailed information.
Background technology
Medical image reaction be that the X line penetrates on the path each physiological tissue position of human body to the accumulated value of X line absorption amount, and each physiological tissue mutually overlaps in the human body, and some institutional frameworks are owing to overlap with the bigger tissue of X line absorption amount and can't clearly show on X line image.CR (Computer Radiography in addition, computed radiography) system is because the laser that the phosphorus particle in the image plate makes the X line exist laser scanner in scattering and the scanning process in the imaging process exists scattering when passing the deep of image plate, thereby make image blurringly, reduced image resolution ratio.Application image enhancement process method highlights organization edge and details, becomes the active demand of Medical Image Processing.
Image enhancement processing method commonly used at present has the window of accent processing method, histogram equalization processing method, fuzzy mask sharpening processing method (Unsharp Masking) etc.Transfer the window processing method can improve the integral image contrast within the specific limits, but the edge enhancing of details and tissue is not had obvious effects; The histogram equalization processing method is calculated the gradation conversion function based on the statistics with histogram result, after the histogram equalization processing, the histogram transformation of image is near equally distributed histogram, to obtain maximum amount of information, it can only obtain the result that overall equalization is handled, and can't realize the effect of the edge enhancing of details and tissue.
Unsharp Masking processing method mainly is made up of three steps:
1. from original image G,,, obtain the low pass that a width of cloth comprises the image low-frequency component and smoothly scheme L image smoothing through low pass filter filtering;
2. deduct low pass with original image G and smoothly scheme L, extract the high fdrequency component H of original image G;
3. superpose with original image G again after high fdrequency component H being multiply by a reinforcing coefficient A,
Generate figure R as a result.
Be described below with mathematical formulae:
R(x，y)＝G(x，y)+A×(G(x，y)-L(x，y))
Wherein, R is the image after strengthening.
When image is carried out low-pass filtering, adopt the method for convolution to realize usually, filter convolution kernel size has determined the space frequency strip that will strengthen.In order to obtain desirable effect, the convolution kernel size should be adjusted to the size of objects.
There is following shortcoming in Unsharp Masking processing method:
1, can't strengthen object on the different space frequency strip simultaneously, objects often appears on the different space frequency strip;
2, when the filter convolution kernel is big, calculating process is consuming time too many, can't satisfy clinical demand.
Summary of the invention
Technical problem to be solved by this invention is: provide a kind of object that strengthens simultaneously on the image different space frequency strip, to highlight the medical image enhancing processing method of organization edge and detailed information.
In order to solve the problems of the technologies described above, the present invention adopts following technical scheme: a kind of medical image enhancing processing method, by low pass filter filtering, to the original image convolution, with image smoothing, obtain the low pass smoothed image, deduct the low pass smoothed image with original image, extract high fdrequency component, the original image that is added to is characterized in that: the image that adopts the convolution kernel convolution of fixed size to dwindle, extract the high fdrequency component on the different space frequency strip, each high fdrequency component is amplified to original image size, and original image simultaneously is added to.
In order to solve the problems of the technologies described above, the present invention adopts following further technical scheme: this method may further comprise the steps:
A) from original image, adopt the fixed size convolution kernel through low pass filter filtering, obtain the low pass figure that a width of cloth comprises the image low-frequency component, deduct low pass figure with original image, obtain the high fdrequency component of original image;
B) on low pass figure horizontal direction and vertical direction, carry out the dot interlace sampling, generate downscaled images; Adopt the fixed size convolution kernel through low pass filter filtering, obtain the low pass figure that a width of cloth comprises the downscaled images low-frequency component, deduct the low pass figure that comprises the downscaled images low-frequency component with downscaled images; Obtain high fdrequency component, high fdrequency component is amplified to the size of original image;
C), with comprising the low pass figure of downscaled images low-frequency component, repeating step b), the rest may be inferred;
D) high fdrequency component of each grade generates figure as a result with the original image stack successively.
Compared with prior art the present invention has following effect:
1, can strengthen object on the image different space frequency strip simultaneously, effectively strengthen the information of image organizational edge and details.
2, adopt the mode of the image that the convolution kernel convolution of fixed size dwindled to replace mode, shortened the spent time of convolution greatly, satisfy clinical demand with different convolution kernel convolution original images.
Description of drawings:
Fig. 1 is a process flow block diagram of the present invention
Fig. 2 is the original image G1 of chest x-ray image
Fig. 3 for wide and high all be half image G2 of G1
Fig. 4 for wide and high all be half image G3 of G2
Fig. 5 for wide and high all be half image G4 of G3
Fig. 6 is the low pass figure L1 of original image G1
Fig. 7 is the low pass figure L2 of image G2
Fig. 8 is the low pass figure L3 of image G3
Fig. 9 is the low pass figure L4 of image G4
Figure 10 is the high fdrequency component H1 of original image G1
Figure 11 is the high fdrequency component H2 of image G2
Figure 12 is the high fdrequency component H3 of image G3
Figure 13 is the high fdrequency component H4 of image G4
Figure 14 is treated image.
Embodiment
Describe principle of the present invention and method in detail below in conjunction with accompanying drawing:
When adopting existing Unsharp Masking method that image is handled, can effectively strengthen and the close object of low pass filter convolution kernel size.Therefore, this image is carried out repeatedly Unsharp Masking handle, use the low pass filter convolution kernel of different sizes respectively, can strengthen the object on the different space frequency strip simultaneously.
Concrete steps are as follows:
1. from original image G, adopt convolution kernel N1, obtain the low pass figure L1 that a width of cloth comprises the image low-frequency component through low pass filter 1 filtering;
2. deduct low pass figure L1 with original image G, can obtain the high fdrequency component H1 of original image G;
3. superpose with original image G again after H1 being multiply by a reinforcing coefficient A1;
4. use convolution kernel N2 instead through low pass filter 2 filtering, reinforcing coefficient A2, repeat the 1st, 2,3 steps;
5.…
I. use convolution kernel Ni instead through low pass filter i filtering, reinforcing coefficient Ai, repeat the 1st, 2,3 steps.
Be described below with mathematical formulae:
R(x，y)＝G(x，y)+A
1×(G(x，y)-L
1(x，y))+……+A
i×(G(x，y)-L
i(x，y))
The method is equal to the Unsharp Masking that has used repeatedly different big or small low pass filter convolution kernels and handles, though this method can solve the shortcoming that single UnsharpMasking method can't strengthen the object on the different space frequency strip simultaneously, but increased frequency with Unsharp Masking processing, convolution kernel also constantly increases, according to the convolution algorithm principle, the convolution kernel participation convolution that size is N for each pixel, needs through following a few step computings:
1.N * N multiplying
2.N * N-1 sub-addition computing
3.1 inferior division arithmetic
Therefore finish square being directly proportional of required time of convolution algorithm and N (convolution kernel), be directly proportional with the size of image, adopt time that this method handles image to handle the summation of holding time greater than Unsharp Masking repeatedly, obviously speed can't satisfy clinical demand.In order to reduce amount of calculation, shorten the time of image processing, the image that the present invention adopts the convolution kernel convolution of fixed size to dwindle, extract the high fdrequency component on the different space frequency strip, each high fdrequency component is amplified to original image size, the original image that is added to simultaneously is to highlight organization edge and detailed information.
With reference to Fig. 1, the present invention specifically is made up of following steps:
1. from original image G1, adopt convolution kernel N1, obtain the low pass figure L1 that a width of cloth comprises original image G1 low-frequency component through low pass filter 1 filtering; Deduct low pass figure L1 with original image G1, obtain the high fdrequency component H1 of original image G1; H1 superposes with original image G1 after be multiply by a reinforcing coefficient A1 again;
2. carry out the dot interlace sampling on low pass figure L1 horizontal direction and the vertical direction, generate downscaled images G2, adopt convolution kernel N1, obtain the low pass figure L2 that a width of cloth comprises image G2 low-frequency component through low pass filter 1 filtering; Deduct low pass figure L2 with figure G2 and obtain high fdrequency component H2; H2 is taken advantage of in reinforcing coefficient A2, be amplified to the size of original image G1 again, superpose with original image G1;
3. on low pass figure L2 horizontal direction and vertical direction, carry out the dot interlace sampling, generate downscaled images G3, adopt convolution kernel N1, obtain the low pass figure L3 that a width of cloth comprises image G3 low-frequency component through low pass filter 1 filtering; Deduct low pass figure L3 with figure G3 and obtain high fdrequency component H3; H3 is taken advantage of in reinforcing coefficient A3, be amplified to the size of original image G1, superpose with original image G1;
4, the rest may be inferred, and repeating step 2, the hihgs of each grade superpose with original image G1 successively, generates figure as a result.
In the above-mentioned steps, the purpose of dot interlace sampling is in order to extract the object on the different space frequency strip, dot interlace can be every a point, two points and even several points, every a point sampling, generate and dwindle 1/2nd image, 1/3rd image is dwindled in the next but two dot generation, dwindle 1/4th image every three dot generation, the rest may be inferred, theoretically, and can be up to the wide and higher primary school that dwindles figure Gi in the size of the convolution kernel of low pass filter.
In the above-mentioned steps, high fdrequency component multiply by a reinforcing coefficient according to the definition of profile and the size of noise, if profile is unintelligible and noise is bigger, reinforcing coefficient can be 0.
Adopt above-mentioned processing method, figure has comprised the high fdrequency component of each grade as a result, has strengthened the organization edge and the detailed information of image.Adopt the mode that image is dwindled simultaneously, the convolution kernel size of each convolution is immobilized, shortened the spent time of convolution greatly, satisfy clinical demand.
Embodiment 1:
As shown in Figure 2, original image G1 is the typical chest x-ray image of a width of cloth.
Picture characteristics is as follows:
Pixel depth: 12 of gray scales;
Picture size: wide by 2560, high by 3072;
Pixel size: 0.139mm * 0.139mm;
The low pass filter of selecting is the gauss low frequency filter of 5 pixels for the convolution kernel size.
Implementation process is as follows:
1. with wide 2560 high 3072 original image G1 convolution kernel 5 gauss low frequency filter filtering, obtain the low pass figure L1 (as shown in Figure 6) that a width of cloth comprises original image G1 low-frequency component, original image G1 deducts low pass figure L1, obtain the high fdrequency component H1 (as shown in figure 10) of original image G1, as can be seen from Fig. 10, high fdrequency component H1 does not have tangible boundary profile information, it mainly is noise, reinforcing coefficient A1 to its application is 0, to suppress the amplification of noise, superposes with original image G1;
2, on low pass figure L1 horizontal direction and vertical direction every a point sampling, generate wide 1280 high 3072 dwindle figure G2 (G2 wide and high all is half of G1 as shown in Figure 3), with the G2 convolution kernel 5 gauss low frequency filter filtering, obtain a width of cloth and comprise the low pass figure L2 (as shown in Figure 7) that dwindles figure G2 low-frequency component, deduct low pass figure L2 with figure G2, obtain the high fdrequency component H2 (as shown in figure 11) of figure G2, H2 has boundary profile more clearly, reinforcing coefficient A2 to its application is 1, superposes with original image G1 after being amplified to original image G1 size again.
3, on low pass figure L2 horizontal direction and vertical direction every a point sampling, generate wide 640 high 768 dwindle figure G3 (G3 wide and high all is half of G2 as shown in Figure 4), with figure G3 convolution kernel 5 gauss low frequency filter filtering, obtain a width of cloth and comprise the low pass figure L3 (as shown in Figure 8) that dwindles figure G3 low-frequency component, deduct low pass figure L3 with figure G3, obtain the high fdrequency component H3 (as shown in figure 12) of G3, H3 has boundary profile more clearly, reinforcing coefficient A3 to its application is 1, superposes with original image G1 after being amplified to original image G1 size again.
4, on low pass figure L3 horizontal direction and vertical direction every a point sampling, generate wide 320 high 384 dwindle figure G4 (G4 wide and high all is half of G3 as shown in Figure 5), with figure G4 convolution kernel 5 gauss low frequency filter filtering, obtain a width of cloth and comprise the low pass figure L4 (as shown in Figure 9) that dwindles figure G4 low-frequency component, deduct low pass figure L4 with figure G4, obtain the high fdrequency component H4 (as shown in figure 13) of figure G4, H4 has boundary profile more clearly, reinforcing coefficient A4 to its application is 1, superposes with original image G1 after being amplified to original image G1 size again.
From Figure 14 as seen, after the employing said method is handled image, be that the details or the edge of bigger tissue of image all obtained strengthening effectively.At CPU P4 1.6G, under the machine environment of internal memory 512M, original image G1 convolution 756ms consuming time, figure G1, G2, G3, the total 1064ms consuming time of G4 convolution, 1.5 seconds with the interior entire image processing process of finishing.
Embodiment 2:
Test machine is identical with embodiment 1, and the low pass filter of selection is the gauss low frequency filter of 7 pixels for the convolution kernel size.Adopt this method, original image G1 convolution 1054ms consuming time, figure G1, G2, G3, the total 2100ms consuming time of G4 convolution.
Embodiment 3
Test machine is identical with embodiment 1, and the low pass filter of selection is the gauss low frequency filter of 9 pixels for the convolution kernel size.Adopt this method, original image G1 convolution 2696ms consuming time, figure G1, G2, G3, the total 3824ms consuming time of G4 convolution.
With the foregoing description as can be known, the convolution kernel size is big more, needs consumed time long more; For time of guaranteeing an image processing in 2 seconds, the size of convolution kernel should be smaller or equal to 5 pixels, but the convolution kernel size is more little, figure will introduce many more noises as a result, so the size of convolution kernel is that 5 pixels are for best.
The present invention can overcome the defective of Unsharp Masking processing method, strengthens the object of different space frequency strip in the conventional medical image in 2 seconds simultaneously, has improved the reinforced effects and the accuracy of image, has bigger social benefit and economic benefit.
Claims (6)
1, a kind of medical image enhancing processing method, by low pass filter filtering, to the original image convolution, with image smoothing, obtain the low pass smoothed image, deduct the low pass smoothed image with original image, extract high fdrequency component, original image is added to, it is characterized in that: the image that the convolution kernel convolution of employing fixed size was dwindled, extract the high fdrequency component on the different space frequency strip, each high fdrequency component is amplified to original image size, original image simultaneously is added to.
2, the described a kind of medical image enhancing processing method of claim 1 is characterized in that this method may further comprise the steps:
A) from original image, adopt the fixed size convolution kernel through low pass filter filtering, obtain the low pass figure that a width of cloth comprises the image low-frequency component, deduct low pass figure with original image, obtain the high fdrequency component of original image;
B) on low pass figure horizontal direction and vertical direction, carry out the dot interlace sampling, generate downscaled images; Adopt the fixed size convolution kernel through low pass filter filtering, obtain the low pass figure that a width of cloth comprises the downscaled images low-frequency component, deduct the low pass figure that comprises the downscaled images low-frequency component with downscaled images, obtain high fdrequency component, high fdrequency component is amplified to the size of original image;
C) with the low pass figure that comprises the downscaled images low-frequency component, repeating step b), the rest may be inferred;
D) high fdrequency component of each grade generates figure as a result with the original image stack successively.
3, a kind of medical image enhancing processing method as claimed in claim 2, it is characterized in that: the described dot interlace of step b) count at least one, dwindle 1/2nd image every a dot generation, the image of next but two dot generation 1/3rd dwindles, dwindle 1/4th every three points, the rest may be inferred.
4, as claim 1 or 2 or 3 described a kind of medical image enhancing processing methods, it is characterized in that: described high fdrequency component multiply by a reinforcing coefficient according to the size of noise.
5, a kind of medical image enhancing processing method as claimed in claim 4 is characterized in that: described convolution kernel is smaller or equal to 5 pixels
6, a kind of medical image enhancing processing method as claimed in claim 5 is characterized in that: described convolution kernel size is 5 pixels.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200610049357 CN1819621A (en) | 2006-01-25 | 2006-01-25 | Medical image enhancing processing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200610049357 CN1819621A (en) | 2006-01-25 | 2006-01-25 | Medical image enhancing processing method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1819621A true CN1819621A (en) | 2006-08-16 |
Family
ID=36919293
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200610049357 Pending CN1819621A (en) | 2006-01-25 | 2006-01-25 | Medical image enhancing processing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1819621A (en) |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102142134A (en) * | 2011-04-29 | 2011-08-03 | 中山大学 | Image detail enhancement method based on three-dimensional grid smoothing model |
| CN102201109A (en) * | 2010-03-26 | 2011-09-28 | 株式会社岛津制作所 | Image processing method and radiographic apparatus using the same |
| CN102238376A (en) * | 2010-04-28 | 2011-11-09 | 鸿富锦精密工业（深圳）有限公司 | Image processing system and method |
| CN101697230B (en) * | 2009-11-03 | 2012-01-04 | 蒋慧琴 | Device and method for adaptive enhancement treatment of medical image |
| CN102332162A (en) * | 2011-09-19 | 2012-01-25 | 西安百利信息科技有限公司 | Method for automatic recognition and stage compression of medical image regions of interest based on artificial neural network |
| CN102696217A (en) * | 2009-09-30 | 2012-09-26 | 夏普株式会社 | Image enlargement device, image enlargement program, and display apparatus |
| CN102737365A (en) * | 2011-03-22 | 2012-10-17 | 株式会社尼康 | Image processing apparatus, imaging apparatus, storage medium storing image processing program, and image processing method |
| CN101815157B (en) * | 2009-02-24 | 2013-01-23 | 虹软(杭州)科技有限公司 | Image and video amplification method and relevant image processing device |
| CN101584204B (en) * | 2006-12-27 | 2013-08-21 | 英特尔公司 | Method and apparatus of content adaptive detailing filtering for digital pictures |
| CN103606135A (en) * | 2013-12-03 | 2014-02-26 | 山东中医药大学 | Medical image enhancement processing method |
| CN103810672A (en) * | 2012-11-12 | 2014-05-21 | 腾讯科技（深圳）有限公司 | Method and device for image fuzzy processing |
| CN106923850A (en) * | 2017-02-22 | 2017-07-07 | 北京唯迈医疗设备有限公司 | The generation method and equipment of a kind of digital subtraction image |
| CN108369732A (en) * | 2015-11-26 | 2018-08-03 | 皇家飞利浦有限公司 | Device with the user interface for enhancing medical image |
| CN108876733A (en) * | 2018-05-30 | 2018-11-23 | 上海联影医疗科技有限公司 | A kind of image enchancing method, device, equipment and storage medium |
| CN110675324A (en) * | 2018-07-02 | 2020-01-10 | 上海寰声智能科技有限公司 | 4K ultra-high definition image sharpening processing method |
| CN110796015A (en) * | 2019-09-27 | 2020-02-14 | 深圳市大拿科技有限公司 | Remote monitoring method and device |
| CN111192208A (en) * | 2019-12-18 | 2020-05-22 | 合肥登特菲医疗设备有限公司 | Tooth CR image enhancement method and device based on side window filter |
| CN114119614A (en) * | 2022-01-27 | 2022-03-01 | 天津风霖物联网科技有限公司 | Method for remotely detecting cracks of building |
-
2006
- 2006-01-25 CN CN 200610049357 patent/CN1819621A/en active Pending
Cited By (27)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101584204B (en) * | 2006-12-27 | 2013-08-21 | 英特尔公司 | Method and apparatus of content adaptive detailing filtering for digital pictures |
| CN101815157B (en) * | 2009-02-24 | 2013-01-23 | 虹软(杭州)科技有限公司 | Image and video amplification method and relevant image processing device |
| CN102696217A (en) * | 2009-09-30 | 2012-09-26 | 夏普株式会社 | Image enlargement device, image enlargement program, and display apparatus |
| CN101697230B (en) * | 2009-11-03 | 2012-01-04 | 蒋慧琴 | Device and method for adaptive enhancement treatment of medical image |
| CN102201109A (en) * | 2010-03-26 | 2011-09-28 | 株式会社岛津制作所 | Image processing method and radiographic apparatus using the same |
| CN102201109B (en) * | 2010-03-26 | 2013-08-28 | 株式会社岛津制作所 | Image processing method and radiographic apparatus using the same |
| CN102238376A (en) * | 2010-04-28 | 2011-11-09 | 鸿富锦精密工业（深圳）有限公司 | Image processing system and method |
| CN102238376B (en) * | 2010-04-28 | 2014-04-23 | 鸿富锦精密工业（深圳）有限公司 | Image processing system and method |
| CN102737365A (en) * | 2011-03-22 | 2012-10-17 | 株式会社尼康 | Image processing apparatus, imaging apparatus, storage medium storing image processing program, and image processing method |
| CN102737365B (en) * | 2011-03-22 | 2016-12-14 | 株式会社尼康 | Image processing apparatus, camera head and image processing method |
| CN102142134A (en) * | 2011-04-29 | 2011-08-03 | 中山大学 | Image detail enhancement method based on three-dimensional grid smoothing model |
| CN102332162A (en) * | 2011-09-19 | 2012-01-25 | 西安百利信息科技有限公司 | Method for automatic recognition and stage compression of medical image regions of interest based on artificial neural network |
| CN103810672A (en) * | 2012-11-12 | 2014-05-21 | 腾讯科技（深圳）有限公司 | Method and device for image fuzzy processing |
| CN103606135A (en) * | 2013-12-03 | 2014-02-26 | 山东中医药大学 | Medical image enhancement processing method |
| CN108369732B (en) * | 2015-11-26 | 2022-08-23 | 皇家飞利浦有限公司 | Device with user interface for enhancing medical images |
| CN108369732A (en) * | 2015-11-26 | 2018-08-03 | 皇家飞利浦有限公司 | Device with the user interface for enhancing medical image |
| CN106923850A (en) * | 2017-02-22 | 2017-07-07 | 北京唯迈医疗设备有限公司 | The generation method and equipment of a kind of digital subtraction image |
| CN108876733A (en) * | 2018-05-30 | 2018-11-23 | 上海联影医疗科技有限公司 | A kind of image enchancing method, device, equipment and storage medium |
| WO2019228481A1 (en) * | 2018-05-30 | 2019-12-05 | Shanghai United Imaging Healthcare Co., Ltd. | Systems and methods for image processing |
| CN108876733B (en) * | 2018-05-30 | 2021-11-09 | 上海联影医疗科技股份有限公司 | Image enhancement method, device, equipment and storage medium |
| US11599982B2 (en) | 2018-05-30 | 2023-03-07 | Shanghai United Imaging Healthcare Co., Ltd. | Systems and methods for image processing |
| CN110675324A (en) * | 2018-07-02 | 2020-01-10 | 上海寰声智能科技有限公司 | 4K ultra-high definition image sharpening processing method |
| CN110675324B (en) * | 2018-07-02 | 2023-10-10 | 上海寰声智能科技有限公司 | 4K ultra-high definition image sharpening processing method |
| CN110796015A (en) * | 2019-09-27 | 2020-02-14 | 深圳市大拿科技有限公司 | Remote monitoring method and device |
| CN111192208A (en) * | 2019-12-18 | 2020-05-22 | 合肥登特菲医疗设备有限公司 | Tooth CR image enhancement method and device based on side window filter |
| CN111192208B (en) * | 2019-12-18 | 2024-04-09 | 合肥登特菲医疗设备有限公司 | Tooth CR image enhancement method and device based on side window filter |
| CN114119614A (en) * | 2022-01-27 | 2022-03-01 | 天津风霖物联网科技有限公司 | Method for remotely detecting cracks of building |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1819621A (en) | Medical image enhancing processing method | |
| CN107481297B (en) | CT image reconstruction method based on convolutional neural network | |
| Huang et al. | CaGAN: A cycle-consistent generative adversarial network with attention for low-dose CT imaging | |
| US7362845B2 (en) | Method and apparatus of global de-noising for cone beam and fan beam CT imaging | |
| CN100342398C (en) | Image processing method and apparatus | |
| CN1934587A (en) | Method, computer program product and apparatus for enhancing a computerized tomography image | |
| CN1889125A (en) | Medical radiation image detail enhancing method | |
| CN1919144A (en) | Ultrasonic image enhancement and spot inhibition method | |
| Wang et al. | Multiscale penalized weighted least-squares sinogram restoration for low-dose X-ray computed tomography | |
| US20100092102A1 (en) | Method and apparatus for image processing | |
| CN101043579A (en) | Method for small detail restoration in digital images | |
| EP1647935A3 (en) | Method and apparatus for enhancing image quality of a two-dimensional ultrasound image | |
| CN1426019A (en) | Method for simutaneously suppressing undershoot and over shoot for increasing digital image | |
| CN1708108A (en) | Apparatus and method to remove jagging artifact | |
| JP2008104761A5 (en) | ||
| CN103186888A (en) | Method and device for removing CT (computed tomography) image noises | |
| CN1917577A (en) | Method of reducing noise for combined images | |
| CN1921562A (en) | Method for image noise reduction based on transforming domain mathematics morphology | |
| CN1517941A (en) | Impact suppession method and device for image detail enhancment | |
| CN104182939A (en) | Medical image detail enhancement method | |
| CN1920881A (en) | Image noise reducing method for Contourlet transform | |
| CN105723416B (en) | Image denoising method | |
| CN101051384A (en) | Self adaptive reinforced digital image reducing blood vessle image | |
| JP4783006B2 (en) | Method and apparatus for region segmentation based image manipulation | |
| CN105469358B (en) | A kind of image processing method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |