CN1904907A - High-speed image matching detecting system and method - Google Patents
High-speed image matching detecting system and method Download PDFInfo
- Publication number
- CN1904907A CN1904907A CN 200510028183 CN200510028183A CN1904907A CN 1904907 A CN1904907 A CN 1904907A CN 200510028183 CN200510028183 CN 200510028183 CN 200510028183 A CN200510028183 A CN 200510028183A CN 1904907 A CN1904907 A CN 1904907A
- Authority
- CN
- China
- Prior art keywords
- image
- high speed
- flush bonding
- matching
- bonding processor
- 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.)
- Granted
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a high speed image matching testing system that uses hardware accelerator to realize FFT and reverse FFT operation, uses embedded processor to operate and control. It discloses the method to use the system to realize high speed image matching testing. It adopts the arithmetic of detecting shift factor and detecting rotating factor in application to realize high speed image matching testing and high accuracy and high throughput process to make image matching detecting more efficient and costs lower.
Description
The present invention relates to a kind of detection system, especially a kind of high-speed image matching detecting system; The invention still further relates to a kind of method of utilizing native system to realize the high speed image coupling.
Background technology
Seldom can use automatic detection device in the industrial at home production quality control, generally all be to adopt manually to detect, and so not only cost is high, and inefficiency.What the pick-up unit that has adopted is the structure of computing machine or industrial computer, also have plenty of the flush bonding processor of employing structure, but cost is higher, and arithmetic speed is all slow, and is subjected to the influence of environment easily, and reliability is low.The algorithm that these devices adopt is a matching by cross correlation, or its arithmetic speed has been carried out some improved algorithms, but nonetheless, still can not reach the real-time requirement that high speed image is handled.
Matching by cross correlation is a most basic a kind of statistical match method.The simple crosscorrelation coupling requires template image to have similar yardstick and half-tone information with image to be matched.Template window travels through on image to be detected, calculates the cross correlation value of each position template image and image counterpart to be detected, and the position of simple crosscorrelation information maximum is the position of template correspondence in image to be detected.
Simple crosscorrelation definition commonly used has following two kinds of forms:
Wherein, ((x y) represents image function to be matched to I to M for x, y) representation template function.
More than two formula calculate is normalized cross-correlation coefficient, though this thinking is simple, along with the increase of image, computing will be very consuming time, for this reason, people have proposed a variety of accelerating algorithm again.
Wherein reasonable have a fast fourier transform relevant matches method.
Fast fourier transform relevant matches method is that image is transformed from a spatial domain to frequency field, and template and image to be detected have converted the complex multiplication operation of frequency spectrum on the frequency field at the computing cross-correlation on the spatial domain.
FM(u，v)＝FFT(M(x，y))
FI(u，v)＝FFT(I(x+i，y+j))
Corr(u，v)＝FM*FI
C(i，j)＝IFFT(Corr(u，v))
Wherein: FM (u, v) and FI (u, v) be respectively M (x, y) and I (x+i, Fourier direct transform y+j).
In frequency domain, the translation of image in spatial domain, rotation, change in size all has corresponding expression part, and this is one of advantage of this method.On the other hand, this method can be got rid of the noise of frequency dependence, thereby obtains the good coupling of robustness.Simultaneously, Fourier transform can be realized by hardware, has improved operation efficiency greatly, speed and data throughout.Now, there have been a lot of fast fourier transform special chips to be applied to improve the performance of system greatly in the various industrial detection systems.
Summary of the invention
Technical matters to be solved by this invention provides a kind of high-speed image matching detecting system, and it can be reached improve the desired processing speed of high speed image matching detection device, and the precision height, real-time, data throughout is big.The present invention also provides a kind of method of utilizing said system, to realize the high speed image matching detection.
For solving the problems of the technologies described above, the technical scheme of a kind of high-speed image matching detecting system of the present invention is to comprise: flush bonding processor is used for the pre-service of image, processing, decision process and system's control of intermediate data; Program storage is connected with described flush bonding processor, is used for storage and operational outfit Control Software, stores processor result; Output display unit is connected with described flush bonding processor, is used for the man-machine interaction of system and the demonstration of result; The data processing hardware accelerator is connected with described flush bonding processor, be used for the collection of image after, the hardware-accelerated processing of data and the control of tupe, and the response flush bonding processor processing requirements; The image recording device according to the instruction that system sends, is used for the high speed acquisition of view data; The video coding device is connected with the data processing hardware accelerator with described image recording device, is used for the image signal decoding and the digitizing of will be gathered; Video memory is connected with described data processing hardware accelerator, is used for the storage of images acquired and the buffer-stored of deal with data.
The present invention utilizes said system to realize that the technical scheme of the method for high speed image matching detection is, comprise the steps: at first enrollment image in system, enter image acquisition matching treatment pattern then, with image recording device images acquired, image to institute's typing is handled, this processing comprises translation processing and rotation processing, compares with template image afterwards, exports the result of matching detection at last.
Further improvements in methods as a kind of high speed image matching detection of the present invention are, the translational movement that in image acquisition matching treatment pattern image is carried out detects to be handled, the steps include: that template image and detected image are after carrying out fast two-dimensional fourier transformation, its spectrum results is carried out separating of amplitude spectrum and phase spectrum and extracted its phase spectrum as research object, phase spectrum to both synthesizes, and then synthetic result carried out two-dimentional inverse fast fourier transform, obtain relevant surfaces, ask for the peak value and the translational movement information of relevant surfaces, correlation peak and translational movement information input decision algorithm are obtained a judgement amount, if it is greater than predefined decision rule, it is relevant then to be judged to be two images, if it is uncorrelated that it, then is judged to be two images less than predefined decision rule.
As another kind of further improvement of the method for a kind of high speed image matching detection of the present invention be, the rotation amount that in image acquisition matching treatment pattern image is carried out detects to be handled, the steps include: that template image and reference picture are after carrying out fast two-dimensional fourier transformation, its spectrum results is carried out separating of amplitude spectrum and phase spectrum and extraction comprises the amplitude spectrum of rotation amount information as research object, amplitude spectrum to both carries out logarithm operation, then its algorithm flow that carries out just entering behind the polar coordinate transform top detection side-play amount is obtained correlation peak and translational movement information, be the translational movement information translation after the rotation amount information and correlation peak input decision algorithm obtains a judgement amount, if it is greater than predefined decision rule, it is relevant then to be judged to be two images, if less than predefined decision rule, it is uncorrelated then to be judged to be two images.
The present invention utilizes hardware system to realize Fourier transform and inverse transformation, and it is high-speed to have realized that images match detects, the processing of high precision and high-throughput, and the efficient that images match is detected is higher, and cost is lower.
Description of drawings
Below in conjunction with drawings and Examples the present invention is further described:
Fig. 1 is the structural drawing of a kind of high-speed image matching detecting system of the present invention;
Fig. 2 is the structural drawing of image recording device of the present invention;
Fig. 3 is the structural drawing of data processing hardware accelerator of the present invention;
Fig. 4 is the process flow diagram of the method for a kind of high speed image matching detection of the present invention;
Fig. 5 is the frame diagram of image translation algorithm of the present invention;
Fig. 6 is the frame diagram of image rotation algorithm of the present invention;
Fig. 7 is the frame diagram of image matching algorithm of the present invention.
Embodiment
In order to improve processing speed, adopted the structure of " flush bonding processor+data processing hardware accelerator " in the native system, realize fast Flourier direct transform/inverse fast fourier transform computing with dedicated hardware accelerators, carry out other computing and system's control with flush bonding processor.So not only improve speed but also had certain dirigibility.In order to improve the precision of images match, in application, adopted and to have detected the algorithm that displacement can detect rotation amount again simultaneously.
The structure of a kind of high-speed image matching detecting system of the present invention comprises as shown in Figure 1:
Data processing hardware accelerator 2 is connected with described flush bonding processor, be used for the collection of image after, the hardware-accelerated processing of data and the control of tupe, and the response flush bonding processor processing requirements.There is the hardware of FFT/IFFT as well as to realize kernel in the described data processing hardware accelerator.
In order to cooperate the flush bonding processor at a high speed and the external bus interface of data processing hardware accelerator, adopt program dynamic storage at a high speed, it has very high bus speed.Video memory also is to adopt static memory at a high speed, and it also has very high bus speed.Can satisfy the requirement of the outside general line interface of built-in processor high speed.
After system powered on, flush bonding processor carried out initial setting up by total demand pairs according to the related register in the processing hardware accelerator, only finish a series of initial setting up after, the data processing hardware accelerator could be started working normally.
As shown in Figure 2, just begin to send out an image typing trigger pip to the image recording device after the data processing hardware accelerator receives image typing trigger pip 4 from the outside, the image recording device receives the image that just begins to gather after the image typing trigger pip in the motion of panel height speed.
After the image recording device collected video image, the picture signal of simulation was just sent into Video Decoder, and after video decoder decodes, the data-signal of generation is deposited in the storer through the data processing hardware accelerator.
In data processing hardware accelerator inside three kinds of bus structure are arranged, as shown in Figure 3.
When the value that flush bonding processor is provided with control register was image typing pattern, the data of image recording device typing directly stored in the storer through the data processing hardware accelerator.
When the value that flush bonding processor is provided with control register was the fast fourier transform tupe, the data processing hardware accelerator can directly carry out data processing to the view data in the storer.
When the value that flush bonding processor is provided with control register was the flush bonding processor tupe, flush bonding processor can carry out various processing to the view data in the storer through the data processing hardware accelerator.
After the data of image recording device typing deposit storer in, the data processing hardware accelerator can be set to the flush bonding processor tupe, flush bonding processor can carry out various image pre-service work by the data in the data processing hardware accelerator reference-to storage.
After flush bonding processor finishes pre-service to view data, just the mode bus in the data processing hardware accelerator may be switched to the fast fourier transform tupe, fast Flourier direct transform module in the data processing hardware accelerator will be carried out two-dimentional fast Flourier direct transform to the data in the storer, the direct transform of two dimension fast Flourier is made up of the direct transform of quadratic one-dimensional fast Flourier, carry out the one dimension line translation earlier, again the data after the line translation are carried out the one dimension rank transformation, at last the data after the conversion are deposited in the storer.
Flush bonding processor switches to the flush bonding processor tupe once more with the bus of data processing hardware accelerator inside, flush bonding processor carries out subsequent processes after some conversion to the data in the storer, extracts phase information and amplitude information and is saved in the storer as template image.
The present invention also comprises a kind of method of using said system to realize the high speed image matching detection, its process flow diagram can be referring to Fig. 4, comprise the steps: at first enrollment image in system, enter image acquisition matching treatment pattern then, with image recording device images acquired, image to institute's typing is handled, and compares with template image, exports the result of matching detection at last.
In image acquisition matching treatment pattern, comprise that the translational movement that image is carried out detects processing, can be referring to Fig. 5, the steps include: that template image and detected image are after carrying out fast two-dimensional fourier transformation, its spectrum results is carried out separating of amplitude spectrum and phase spectrum and extracted its phase spectrum as research object, phase spectrum to both synthesizes, and then synthetic result is carried out two-dimentional inverse fast fourier transform just can obtain relevant surfaces (being related function), ask for the peak value (being the maximal value of related function) and the translational movement information of relevant surfaces, correlation peak and translational movement information input decision algorithm are obtained a judgement amount, if it is greater than predefined decision rule, it is relevant then to be judged to be two images, if it is uncorrelated that it, then is judged to be two images less than predefined decision rule.The relevant then pairing coordinate of peak value of two width of cloth images is the relative displacement of two images.Certainly, in the application system of reality, because the content of the precision of computing and actual two images is inconsistent, the related function of two images can not be the unit impact response function.
In image acquisition matching treatment pattern, comprise that the rotation amount that image is carried out detects processing, can be referring to Fig. 6, the steps include: that template image and reference picture are after carrying out fast two-dimensional fourier transformation, its spectrum results is carried out separating of amplitude spectrum and phase spectrum and extraction comprises the amplitude spectrum of rotation amount information as research object, amplitude spectrum to both carries out logarithm operation, and the computing of taking the logarithm is in order to reduce its dynamic range to amplitude spectrum.Then its algorithm flow that carries out just entering behind the polar coordinate transform top detection side-play amount is obtained correlation peak and translational movement information, be the translational movement information translation after the rotation amount information and correlation peak input decision algorithm obtains a judgement amount, equally, if correlation peak is greater than predefined decision rule, it is relevant then to be judged to be two images, if less than predefined decision rule, it is uncorrelated then to be judged to be two images.There is certain corresponding relation in the relevant then pairing coordinate of peak value of two images with the relative rotation amount of two images.
When existing translation has rotation again between two images, the algorithm that detects side-play amount can be combined with the algorithm that detects rotation amount, adopt structural framing shown in Figure 7 to reach the purpose of detection.Do the algorithm that detects rotation amount earlier and try to achieve the anglec of rotation, do the algorithm process that detects side-play amount after again the image of importing being gone back to by this angle.Also can carry out translation to image earlier and handle, and then image is rotated processing.
In addition, system also is provided with Man Machine Interface, can carry out certain operations according to user's requirement.
According to user's requirement, if the translation information between detected image then enters the algorithm that detects translation, carry out the synthetic processing of the phase information of the phase information of template image and detected image to be matched, result is stored in the storer.
According to user's requirement, if the rotation amount information between detected image then enters the algorithm that detects rotation amount, carry out the synthetic processing of the amplitude information of the amplitude information of template image and detected image to be matched, result is stored in the storer.
Flush bonding processor switches to the fast fourier transform tupe once more with the bus of data processing hardware accelerator inside, and the inverse fast fourier transform processing module in the data processing hardware accelerator is made inverse fast fourier transform to the view data in the storer.
Flush bonding processor switches to the flush bonding processor tupe with the bus of data processing hardware accelerator inside, the view data of flush bonding processor after to inverse fast fourier transform carried out decision process, calculate maximum offset and matching value, judge according to the decision rule of setting whether the image that captures and the template image of appointment mate, and the output court verdict.
Claims (10)
1. a high-speed image matching detecting system is characterized in that, comprising:
Flush bonding processor is used for the pre-service of image, processing, decision process and system's control of intermediate data;
Program storage is connected with described flush bonding processor, is used for storage and operational outfit Control Software, stores processor result;
Output display unit is connected with described flush bonding processor, is used for the man-machine interaction of system and the demonstration of result;
The data processing hardware accelerator is connected with described flush bonding processor, be used for the collection of image after, the hardware-accelerated processing of data and the control of tupe, and the response flush bonding processor processing requirements;
The image recording device according to the instruction that system sends, is used for the high speed acquisition of view data;
The video coding device is connected with the data processing hardware accelerator with described image recording device, is used for the image signal decoding and the digitizing of will be gathered;
Video memory is connected with described data processing hardware accelerator, is used for the storage of images acquired and the buffer-stored of deal with data.
2. a kind of high-speed image matching detecting system according to claim 1 is characterized in that, described flush bonding processor has internal multiplier and internal floating point calculation coprocessor, and 5 stage pipeline structure.
3. a kind of high-speed image matching detecting system according to claim 1 is characterized in that, has the hardware of FFT/IFFT as well as to realize kernel in the described data processing hardware accelerator.
4. a kind of high-speed image matching detecting system according to claim 1 is characterized in that, described program storage is the high speed dynamic storage.
5. a kind of high-speed image matching detecting system according to claim 1 is characterized in that, described video memory is a highspeed static memory.
6. method that realizes the high speed image matching detection with the system as claimed in claim 1, it is characterized in that, comprise the steps: at first enrollment image in system, enter image acquisition matching treatment pattern then, with image recording device images acquired, the image of institute's typing to be handled, this processing comprises translation processing and rotation processing, compare with template image afterwards, export the result of matching detection at last.
7. the method for a kind of high speed image matching detection according to claim 6 is characterized in that, in image acquisition matching treatment pattern, earlier image is rotated processing, and then image is carried out translation handle.
8. the method for a kind of high speed image matching detection according to claim 6 is characterized in that, in image acquisition matching treatment pattern, earlier image is carried out translation and handles, and then image is rotated processing.
9. according to claim 6, the method of 7 or 8 described a kind of high speed image matching detection, it is characterized in that, the translational movement that in image acquisition matching treatment pattern image is carried out detects to be handled, the steps include: that template image and detected image are after carrying out fast two-dimensional fourier transformation, its spectrum results is carried out separating of amplitude spectrum and phase spectrum and extracted its phase spectrum as research object, phase spectrum to both synthesizes, and then synthetic result carried out two-dimentional inverse fast fourier transform, obtain relevant surfaces, ask for the peak value and the translational movement information of relevant surfaces, correlation peak and translational movement information input decision algorithm are obtained a judgement amount, if it is greater than predefined decision rule, it is relevant then to be judged to be two images, uncorrelated if it, then is judged to be two images less than predefined decision rule.
10. according to claim 6, the method of 7 or 8 described a kind of high speed image matching detection, it is characterized in that, the rotation amount that in image acquisition matching treatment pattern image is carried out detects to be handled, the steps include: that template image and reference picture are after carrying out fast two-dimensional fourier transformation, its spectrum results is carried out separating of amplitude spectrum and phase spectrum and extraction comprises the amplitude spectrum of rotation amount information as research object, amplitude spectrum to both carries out logarithm operation, then its algorithm flow that carries out just entering behind the polar coordinate transform top detection side-play amount is obtained correlation peak and translational movement information, be the translational movement information translation after the rotation amount information and correlation peak input decision algorithm obtains a judgement amount, if correlation peak is greater than predefined decision rule, it is relevant then to be judged to be two images, if less than predefined decision rule, it is uncorrelated then to be judged to be two images.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100281839A CN100416600C (en) | 2005-07-27 | 2005-07-27 | High-speed image matching detecting system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100281839A CN100416600C (en) | 2005-07-27 | 2005-07-27 | High-speed image matching detecting system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1904907A true CN1904907A (en) | 2007-01-31 |
| CN100416600C CN100416600C (en) | 2008-09-03 |
Family
ID=37674167
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005100281839A Active CN100416600C (en) | 2005-07-27 | 2005-07-27 | High-speed image matching detecting system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100416600C (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101236601B (en) * | 2008-03-11 | 2010-10-06 | 马磊 | Image recognition accelerator and MPU chip possessing image recognition accelerator |
| CN102184521A (en) * | 2011-03-24 | 2011-09-14 | 苏州迪吉特电子科技有限公司 | High-performance image processing system and image processing method |
| CN103885587A (en) * | 2014-02-21 | 2014-06-25 | 联想(北京)有限公司 | Information processing method and electronic equipment |
| WO2016141803A1 (en) * | 2015-03-06 | 2016-09-15 | 华为技术有限公司 | Image recognition accelerator, terminal device and image recognition method |
| CN107146245A (en) * | 2017-05-05 | 2017-09-08 | 北京京东尚科信息技术有限公司 | image matching method and device |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5634092A (en) * | 1994-09-30 | 1997-05-27 | Apple Computer, Inc. | Color image processing system which provides multiple image processing operations through a single interface |
| US6041140A (en) * | 1994-10-04 | 2000-03-21 | Synthonics, Incorporated | Apparatus for interactive image correlation for three dimensional image production |
| CN1184583C (en) * | 2002-05-16 | 2005-01-12 | 王巍 | Method and device for vertifying handwriting |
-
2005
- 2005-07-27 CN CNB2005100281839A patent/CN100416600C/en active Active
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101236601B (en) * | 2008-03-11 | 2010-10-06 | 马磊 | Image recognition accelerator and MPU chip possessing image recognition accelerator |
| CN102184521A (en) * | 2011-03-24 | 2011-09-14 | 苏州迪吉特电子科技有限公司 | High-performance image processing system and image processing method |
| CN102184521B (en) * | 2011-03-24 | 2013-03-06 | 苏州迪吉特电子科技有限公司 | High-performance image processing system and image processing method |
| CN103885587A (en) * | 2014-02-21 | 2014-06-25 | 联想(北京)有限公司 | Information processing method and electronic equipment |
| WO2016141803A1 (en) * | 2015-03-06 | 2016-09-15 | 华为技术有限公司 | Image recognition accelerator, terminal device and image recognition method |
| CN105989352A (en) * | 2015-03-06 | 2016-10-05 | 华为技术有限公司 | Image identification accelerator, terminal device and image identification method |
| CN107851175A (en) * | 2015-03-06 | 2018-03-27 | 华为技术有限公司 | Image recognition accelerator, terminal device and image-recognizing method |
| US10346701B2 (en) | 2015-03-06 | 2019-07-09 | Huawei Technologies Co., Ltd. | Image recognition accelerator, terminal device, and image recognition method |
| CN105989352B (en) * | 2015-03-06 | 2019-08-20 | 华为技术有限公司 | Image recognition accelerator, terminal device and image-recognizing method |
| CN107851175B (en) * | 2015-03-06 | 2020-07-14 | 华为技术有限公司 | Image recognition accelerator, terminal device and image recognition method |
| CN107146245A (en) * | 2017-05-05 | 2017-09-08 | 北京京东尚科信息技术有限公司 | image matching method and device |
| CN107146245B (en) * | 2017-05-05 | 2020-06-05 | 天津京东深拓机器人科技有限公司 | Image matching method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100416600C (en) | 2008-09-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cristianini et al. | On kernel-target alignment | |
| CN1904907A (en) | High-speed image matching detecting system and method | |
| WO2021036642A1 (en) | Intelligent health prediction method and apparatus for electronic device based on digital twin model | |
| CN101079106A (en) | Different fingerprint sensor image information compatible fingerprint identification method | |
| CN1167959A (en) | Method and apparatus for multi-resolution image searching | |
| CN102789633B (en) | Based on the image noise reduction system and method for K-SVD and locally linear embedding | |
| CN1725246A (en) | A kind of human body posture deforming method based on video content | |
| CN1916933A (en) | Video direct reading type automatic meter reading system, and image processing method | |
| CN104933441B (en) | Object detection system and method | |
| CN102208033B (en) | Data clustering-based robust scale invariant feature transform (SIFT) feature matching method | |
| CN1664755A (en) | Video recognition input system | |
| CN103559697A (en) | Scrap paper lengthwise cutting splicing and recovering algorithm based on FFT | |
| CN100342392C (en) | Coarse positioning method for remote sensing image based on Fourier-Mellin transformation | |
| CN1253774C (en) | Positioning system and method for detecting coordinates of vibration source position | |
| CN102842133A (en) | Partial characteristic description method | |
| CN1436424A (en) | Adaptive early exit techniques for image correlation minimum distortion calculation | |
| CN1298162A (en) | Method for analysing digital image texture structure | |
| Kang et al. | Image registration based on harris corner and mutual information | |
| Lang et al. | Integral image based fast algorithm for two-dimensional Otsu thresholding | |
| CN100352282C (en) | Adaptive early exit techniques for image correlation minimum distortion calculation | |
| CN110309689B (en) | Gabor domain gesture recognition detection method based on ultra-wideband radar | |
| CN103278332B (en) | The engine startup detection method to cylinder misfire | |
| CN1873656A (en) | Detection method of natural target in robot vision navigation | |
| CN1845175A (en) | Grain surface damage detecting method based on wavelet and co-occurrence matrix | |
| CN1436426A (en) | Adaptive early exit techniques for image correlation minimum distortion calculation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C56 | Change in the name or address of the patentee |
Owner name: SHANGHAI BWAVE TECHNOLOGY CO., LTD Free format text: FORMER NAME: SHANGHAI BWAVETECH CORPORATION |
|
| CP03 | Change of name, title or address |
Address after: Shanghai city 201203 Keyuan Road, Zhangjiang hi tech park, building 399, floor 6 Patentee after: Shanghai Bwave Technology Co., Ltd. Address before: 201203 Shanghai Zhangjiang hi tech Park Chunxiao Road No. 439 Building No. 2 Patentee before: Shanghai BwaveTech Corporation |