CN104835177A - Star point segmentation method under interference of moonlight and FPGA device of star point segmentation method - Google Patents
Star point segmentation method under interference of moonlight and FPGA device of star point segmentation method Download PDFInfo
- Publication number
- CN104835177A CN104835177A CN201510288957.5A CN201510288957A CN104835177A CN 104835177 A CN104835177 A CN 104835177A CN 201510288957 A CN201510288957 A CN 201510288957A CN 104835177 A CN104835177 A CN 104835177A
- Authority
- CN
- China
- Prior art keywords
- window
- asterism
- module
- adaptive threshold
- gradient
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a star point segmentation method under interference of moonlight and an FPGA device of the star point segmentation method. The method comprises that the local gradient G of a star point window image under the interference of moonlight is calculated; influence of moonlight noise is eliminated via a gradient threshold; and a self-adaptive threshold is used to determine whether the central pixel gray value of a window is higher than the self-adaptive threshold, and if it is determined that a star point or part of the star point in contained in the window and the central pixel gray value of the window is higher than the self-adaptive threshold, the central pixel of the window is determined to be the star point finally. The FPGA device of the method comprises a gray degree window generation module, a local gradient module, a self-adaptive threshold module and a segmentation result processing module. All of star point segmentation is realized by the FPGA device, the power consumption is low, the device is small, system design is concise, the advantages of parallel assembly line calculation are fully played, and the requirement of instantaneity is satisfied.
Description
Technical field
The present invention relates to the lower asterism dividing method of a kind of moonlight interference and FPGA implement device thereof, belong to star sensor technical field.
Background technology
Star sensor, is carry out mating based on asterism information in current field and star chart and obtain the space precise instrument of spacecraft three-axis attitude information, is mainly used in spacecraft celestial navigation.The star sensor course of work mainly comprises three processes: asterism extraction, importance in star map recognition and Attitude Calculation.Asterism is split, it is the first step during asterism extracts, also be basis and a step of key in Star-Sensor Design, its segmentation precision will badly influence the efficiency and precision of subsequent step, and then affects the reliability of star sensor, measuring accuracy and data updating rate etc.
Without in moon star chart, asterism and background are distinguished comparatively large in gray scale, and an available threshold value is effectively separated by the two.Traditional global threshold method has based on histogrammic global threshold method, based on the global threshold method of cluster, based on the global threshold method etc. of entropy.In asterism segmentation, conventional global threshold is that image average adds 3 to 5 times of variances.When causing uneven illumination even when being subject to moon interference, global threshold cannot be partitioned into asterism, now needs according to local image characteristic for each pixel calculates an adaptive threshold.Local threshold method for each pixel calculates a threshold value, can suppress the even interference brought of uneven illumination on the basis of pixel surrounding pixel information.Traditional local threshold method has the method based on local variance, the method based on local contrast, Threshold Surface matching, image block threshold value, method etc. in conjunction with shape information.Tradition local threshold method is applicable to the segmentation of black character under white background, when being applied to the segmentation of the white asterism under black background, although the even impact on segmentation result of uneven illumination can be suppressed, but still the noise that can be partitioned into the moon and be produced by the moon, and faint star point cannot be extracted.In star-fields segmentation, local mean value can be adopted to reinforce the method determination adaptive threshold of definite value, but the method can be partitioned into a moon bright limb, and faint star point cannot be partitioned into.
Because star sensor requires higher to data updating rate, therefore the real-time implementation of star-fields segmentation is extremely important.In order to ensure the real-time of star-fields segmentation, then need fast throughput and the data throughput capabilities of raising system.FPGA is parallel processing structure, can carry out computing of different nature simultaneously, can complete a complex calculation rapidly in a clock, therefore adopts FPGA to realize star-fields segmentation method, can improve system processing speed, requirement of real time.At present, the moon interference under star chart segmentation in research less.
Summary of the invention
The technology of the present invention is dealt with problems: overcome prior art deficiency, a kind of moonlight is provided to disturb lower asterism dividing method and FPGA implement device thereof mainly to solve following a few subproblem: (1) is for the segmentation of the lower asterism of moon interference, current research is less, and traditional threshold segmentation method, the accurate segmentation of the lower asterism of moon interference cannot be applicable to.(2) when having the moon in the star chart obtained and the moon exists sudden change in gray scale, the existing dividing method for the lower asterism of moon interference, will extract the edge of the moon, cannot suppress the interference of the moon completely.(3) the existing dividing method for the lower asterism of moon interference, mainly for be the segmentation of bright star point in star chart, the faint star point in star chart cannot be partitioned into, and split sufficiently complete.(4) because star sensor requires higher to data updating rate, need elevator system processing speed with requirement of real time.
The technology of the present invention solution: the lower asterism dividing method of a kind of moonlight interference, performing step is as follows:
(1) calculate the partial gradient G of the asterism video in window under moonlight interference, described partial gradient G is the maximal value of Grad around asterism video in window pixel, that is:
G(x,y)=max(g)
Wherein g is pixel (x, y) Grad around;
(2) on the basis of partial gradient G gradient, got rid of the impact of moon optical noise by Grads threshold, described Grads threshold is:
l=0.125×m
2(x,y)+210
Wherein m (x, y) is pixel (x, y) image average around, if partial gradient G is greater than Grads threshold l, then illustrates in video in window have asterism or part asterism;
(3) determining to comprise in window on the basis of asterism or part asterism, adopt adaptive threshold to be used for determining whether window center pixel is asterism, if center pixel gray-scale value is higher than adaptive threshold, then center pixel is asterism, described adaptive threshold:
Wherein midvalue is gray-scale value intermediate value, and m (x, y) is local mean value.
Realize a FPGA device for the lower asterism dividing method of moonlight interference, comprising: greylevel window generation module, partial gradient module, adaptive threshold module and segmentation result processing module; Wherein:
Described greylevel window generation module, generates the greylevel window data of different size after buffer memory part gradation data, enter partial gradient module and adaptive threshold module respectively; The result of partial gradient module and adaptive threshold module enters segmentation result processing module, if comprise a part for asterism or asterism in partial gradient module determination window, the window center of adaptive threshold module determination simultaneously grey scale pixel value is higher than adaptive threshold, then window center pixel is finally defined as asterism.
Described partial gradient module comprises Sobel wave filter, gradient window generation module, maximum filter, mean filter, Grads threshold computing module and Grads threshold comparison module; Greylevel window data enter Sobel wave filter computed image gradient, and image gradient generates gradient window data and enters maximum filter after gradient window generation module buffer memory, obtains partial gradient; Meanwhile, greylevel window data enter mean filter and calculate local mean value, and local mean value enters Grads threshold computing module and calculates Grads threshold; The partial gradient calculated and Grads threshold enter Grads threshold comparison module, by comparing the part determining whether to comprise asterism or asterism in window.
Described adaptive threshold module comprises mean filter, median filter, adaptive threshold computing module and adaptive threshold comparison module; Greylevel window data enter mean filter and median filter calculates local mean value and local intermediate value, the local mean value obtained and local intermediate value enter adaptive threshold computing module, calculate adaptive threshold, the adaptive threshold obtained enters adaptive threshold comparison module, by comparing with window center gray-scale value, determine that whether window center pixel is higher than adaptive threshold.
The maximum filter of described gradient window data is separable, can the filtering of rank of advanced units maximal value, then carries out the filtering of row maximal value.
Described mean filter adopts two one dimension mean filter combinations to realize two-dimentional mean filter, namely when new row enter asterism video in window, calculate new row and, and be added with gray scale summation in video in window, then cut the 8th row leaving video in window row and, only need a sub-addition and a subtraction to realize mean filter, and without the need to considering image window size, each row and need use twice, be once it when entering window, be once that it is when leaving window; In order to arrange and second time use, adopt FIFO come buffer memory row and.
The present invention's advantage is compared with prior art:
(1) make full use of the feature of the imaging of the star sensor moon and photodetector digital imagery, the interference of the moon can be suppressed completely.
(2) make full use of the feature of the imaging of star sensor fixed star and photodetector digital imagery, the accurate segmentation of faint star and bright star can be realized.
(3) make full use of the advantage of FPGA in the design of parallel pipeline operation architecture, employing parallel pipeline architecture organizes the computing module in asterism partitioning algorithm, significantly improves arithmetic speed, reduces system process required time.
In a word, the present invention proposes the lower accurate dividing method of asterism of moonlight interference and FPGA implement device thereof, has taken into full account the feature of star sensor fixed star and moon imaging and photodetector digital imagery, has been made up of partial gradient and adaptive threshold two parts.The all working that asterism is split by the present invention all adopts FPGA to realize, and low in energy consumption, miniaturization, system are succinct, can give full play to its parallel pipelining process line computation advantage, requirement of real time.
Accompanying drawing explanation
Fig. 1 is four kinds of image types in the lower star chart of moonlight interference: ground unrest, asterism, the moon optical noise and the moon;
Fig. 2 asterism position and non-asterism position partial gradient, wherein square frame is asterism position gradient, and circle is non-asterism position gradient, and dotted line is matching gained threshold value;
Comprise two kinds of situations of asterism in Fig. 3 window: a center pixel is asterism, b center pixel is not asterism;
The hard-wired overall framework figure of Fig. 4 the inventive method;
Fig. 5 a window operation, wherein black bars is window center pixel, the hardware configuration that b window generates;
Fig. 6 partial gradient function structure chart;
Fig. 7 mean filter Recursive Implementation mode;
Fig. 8 adaptive threshold function structure chart;
Fig. 9 is to the suppression result of the moon, and wherein a is source images, and b is the result after segmentation;
Figure 10 is to the segmentation result of asterism, and wherein a is source images, and b is the result after segmentation;
Figure 11 is the root-mean-square error that the present invention extracts different magnitude asterism segmentation result barycenter;
Figure 12 is under the impact of different brackets Gaussian noise, the root-mean-square error that the present invention extracts asterism segmentation result barycenter.
Embodiment
Below in conjunction with drawings and Examples, the present invention is described in detail.
What the present invention be directed to is the accurate segmentation that the moon disturbs lower star chart, and object is the interference suppressing the moon and noise in star chart completely, completes the accurate segmentation of faint star and bright star.
1. partial gradient
By finding the in-depth analysis of the lower star chart of moonlight interference, it mainly comprises four class video in windows, namely ground unrest, asterism, the moon optical noise and the moon, as shown in Figure 1.Due to asterism and the moon optical noise have obvious grey scale change, therefore image gradient is large, and ground unrest and moon grey scale change slow, image gradient is little.The moon, optical noise was near the moon simultaneously, and local mean value is higher.Therefore, asterism video in window and other three classes video in window can be made a distinction with exclusive PCR.
Gradient can represent intensity and the direction at edge, (x, y) position in piece image f, and it is defined as:
This vector indicates the direction of f at the maximum rate of change at position (x, y) place.Because image is digital quantity, therefore the gradient of image is its digital approximation about partial derivative on neighbourhood of a point, that is:
Consider that FPGA has very high efficiency when carrying out template computing, use the Sobel template of 3 × 3 to obtain partial derivative and be similar to more accurately:
Wherein z
1, z
2, z
3be three pixels of the first row in 3 × 3Sobel template, z
4, z
5, z
6be three pixels of the second row, z
7, z
8, z
9for three pixels of the third line.
Namely come approximate x direction by the difference of the third line and the first row in image 3 × 3 region reciprocal, it is reciprocal that the 3rd row and the difference of first row come approximate y direction.Vector
size be expressed as g (x, y), that is:
It is the value of gradient vector direction rate of change.Because quadratic sum square root calculated amount is comparatively large, usually use absolute value approximate gradient amplitude:
g(x,y)≈|g
x|+|g
y|
Partial gradient is defined as the maximal value of Grad around it, that is:
G(x,y)=max(g)
Wherein g is pixel (x, y) Grad around.
Fig. 2 gives diverse location partial gradient in star chart.Because some non-asterism position also exists larger partial gradient, only utilize partial gradient asterism cannot be distinguished.Therefore need to determine that Grads threshold is to get rid of the impact of moon optical noise adaptively according to local image characteristic.By the partial gradient separatrix of matching asterism and non-asterism position, consider hard-wired complicacy, the Grads threshold obtained is simultaneously:
l=0.125×m
2(x,y)+210
Wherein m (x, y) is pixel (x, y) image average around.If partial gradient G is greater than Grads threshold l, then illustrate in video in window have asterism or part asterism
2. adaptive threshold
Fig. 3 gives in window the schematic diagram of the two kinds of situations comprising asterism.While comprising asterism or part asterism in partial gradient determination window, adaptive threshold is used for determining that whether window center grey scale pixel value is higher than adaptive threshold.If center pixel gray-scale value is higher than adaptive threshold, then center pixel is asterism, as shown in a in Fig. 3, otherwise not, as shown in the b in Fig. 3.Consider hard-wired complicacy, propose a kind of new adaptive threshold:
Wherein midvalue is gray-scale value intermediate value, and m (x, y) is local mean value.In this way, only need calculate intermediate value and local mean value can calculate adaptive threshold, avoid the complex calculation such as division and square root, thus be conducive to FPGA effective implemention.
3.FPGA implement device
Asterism dividing method implement device comprises greylevel window generation module, partial gradient, and adaptive threshold and segmentation result processing module, hardware implementing overall framework as shown in Figure 4.Due to the concurrency between partial gradient and adaptive threshold module, employing FPGA parallel pipeline structure can elevator system processing speed.Because two modules process simultaneously, system processing time is by wherein the longest module decision consuming time.Two inside modules, comprise again several little module, they are processed by parallel pipeline structure, to reduce system cloud gray model required time.Detail is as follows.
Window operation uses window or neighborhood of pixels to carry out operating to obtain result, and it is widely used in image filtering, image convolution etc.In this implement device, window generation module comprises two parts, i.e. greylevel window generation module and gradient window generation module.In greylevel window generation module, asterism dividing method adopts the window of 8 × 8 sizes, and by from left to right, the mode of sliding from top to bottom, carries out mean filter and medium filtering, as shown in a in Fig. 5, adopts the window of 3 × 3 sizes to carry out Sobel filtering simultaneously.Because mean filter needs to use division, and for 8 × 8 window, substitute division by shift operation, thus reduce hardware implementing complexity.The first in first out buffer (FIFO) that asterism dividing method adopts 8 sizes equal with picture traverse come buffer memory 8 row with data-stream form one by one the view data that enters of clock with the window generating 8 × 8, and generate the window of 3 × 3 to the third line by the first row, as shown in the b in Fig. 5.Control module produces read/write signal with the inflow of data in control FIFO and reading.In gradient window generation module, asterism dividing method adopts the window of 8 × 8 sizes, and by from left to right, the mode of sliding from top to bottom, carries out maximal value filtering, and in its 8 × 8 window generating mode greylevel window generation module, 8 × 8 window generating modes are identical.
Partial gradient module comprises Sobel wave filter, gradient window generation module, maximum filter, mean filter, Grads threshold computing module and Grads threshold comparison module, and they are by parallel pipeline structure process, and its hardware structure diagram as shown in Figure 6.In partial gradient module, the greylevel window data of 3 × 3 obtain image gradient value by Sobel wave filter, then generated the gradient window data of 8 × 8 by gradient window generation module, then obtain partial gradient respectively by row maximum filter and row maximum filter.Simultaneously, the greylevel window data of 8 × 8 obtain local mean value by mean filter, then obtain Grads threshold by Grads threshold computing module, the partial gradient of acquisition and Grads threshold enter Grads threshold comparison module by comparing the part determining whether to comprise asterism or asterism in window.Mean filter due to square window is separable, and available two one dimension mean filters combination realizes two-dimentional mean filter.Because weights are equal, make it can Recursive Implementation.When new row enter window, row and computing module calculate its row with and with the 8th row arranged leaving window of buffer memory in FIFO with subtract each other, then be added with gray scale summation in the window with buffer memory in buffer memory, namely finally the summation obtained moved to right 6 obtains local mean value.The Recursive Implementation mode of mean filter as shown in Figure 7.Therefore, only need a sub-addition and a subtraction to realize mean filter, and without the need to considering window size.Each row and need use twice, be once it when entering window, be once that it is when leaving window.In order to arrange and second time use, adopt FIFO come buffer memory row and.
Adaptive threshold module comprises mean filter, median filter, and adaptive threshold computing module and adaptive threshold comparison module, its hardware structure diagram as shown in Figure 8.In adaptive threshold module, the greylevel window data of 8 × 8 obtain local intermediate value respectively by column mean wave filter and row median filter, and meanwhile, the window data of 8 × 8 obtains local mean value by mean filter.The local intermediate value obtained and local mean value obtain adaptive threshold by adaptive threshold computing module.Finally, in threshold value comparison module, the adaptive threshold of acquisition and window center pixel grey scale are compared and determine that whether adaptive threshold is higher than window center gray-scale value.Because mean filter in this module is identical with mean filter in partial gradient module, two modules share mean data.Compare in a large number and swap operation because medium filtering comprises, be difficult to intermediate value in calculation window.Meanwhile, medium filtering is inseparable, cannot realize medium filtering with two one-dimensional filtering devices.Consider hardware implementing, available column median filter adds row median filter and replaces median filter.The intermediate value obtained like this and actual intermediate value closely, can not produce considerable influence to precision.
In segmentation result processing module, the result of partial gradient module and adaptive threshold module enters segmentation result processing module, if comprise a part for asterism or asterism in partial gradient module determination window, the window center of adaptive threshold module determination simultaneously pixel is higher than adaptive threshold, then window center pixel is finally defined as asterism.
The present invention proposes the lower accurate dividing method of asterism of moonlight interference and FPGA implement device thereof, has taken into full account the feature of star sensor fixed star and moon imaging and photodetector digital imagery, has been made up of partial gradient and adaptive threshold two parts.The all working that asterism is split by the present invention all adopts FPGA to realize, and low in energy consumption, miniaturization, system are succinct, can give full play to its parallel pipelining process line computation advantage, requirement of real time.
Embodiment
Adopt the present invention can process the star chart of degenerating because of moon high light and noise effect, suppress the impact of the moon completely, be partitioned into bright star point and faint star point accurately.In order to verify the accuracy that this dividing method is split asterism, for the actual star chart under moon interference and emulation star chart two class image, PC and FPGA carries out a series of emulation testing.The particular content of test is as follows:
In order to verify that the present invention is to the integrality of asterism segmentation in actual star chart and the rejection ability to the moon, shows respectively by the moon suddenlyd change in star chart and all kinds of asterism.As shown in Figure 9, wherein a is source images to the result finally obtained, and b is segmentation result of the present invention.Can find out according to result, the present invention while suppressing the moon and moonlight noise effect completely, accurately can be partitioned into faint star point and bright star point, remains the integrality of asterism, accurately extracts asterism positional information to be conducive to subsequent step.
By using Verilog language, the inventive method is successfully realized on the FPGA of Xilinx company.The the highest of FPGA hardware implementing display can running frequency be 189.609MHz, and the time needed for gray-scale map processing 2048 × 2048 sizes under this frequency is 22.22ms.
In order to verify the accuracy that the present invention is split different magnitude asterism, generating the emulation star chart of a series of different magnitude, comprising totally 8 groups of images such as 0 ~ 6.5 star such as grade, often organize image and comprise 40 asterisms.A series of star-fields segmentation is tested to have utilized the present invention to do on FPGA.Figure 11 is that the result square weighting centroid method that different magnitude star chart obtains on FPGA and PC carries out the contrast that barycenter extracts the square error obtained.Can find out that the present invention is higher to different magnitude asterism segmentation precision, FPGA also can reach identical precision.Figure 12 is under different Gaussian noise, the accuracy comparison that the barycenter that FPGA and PC obtains extracts.Can find out that, under different brackets noise effect, the present invention still can reach very high segmentation precision.Even if FPGA exists a large amount of being similar in realizing, segmentation precision closely when realizing with PC still can be obtained.
There is provided above embodiment to be only used to describe object of the present invention, and do not really want to limit the scope of the invention.Scope of the present invention is defined by the following claims.Do not depart from spirit of the present invention and principle and the various equivalent substitutions and modifications made, all should contain within the scope of the present invention.
Claims (6)
1. the lower asterism dividing method of moonlight interference, is characterized in that performing step is as follows:
(1) calculate the partial gradient G of the asterism video in window under moonlight interference, described partial gradient G is the maximal value of Grad around asterism video in window pixel, that is:
G(x,y)=max(g)
Wherein g is pixel (x, y) Grad around;
(2) on the basis of partial gradient G gradient, got rid of the impact of moon optical noise by Grads threshold, described Grads threshold is:
l=0.125×m
2(x,y)+210
Wherein m (x, y) is pixel (x, y) image average around, if partial gradient G is greater than Grads threshold l, then illustrates in video in window have asterism or part asterism;
(3) determining to comprise in window on the basis of asterism or part asterism, adopt adaptive threshold to be used for determining whether window center pixel is asterism, if center pixel gray-scale value is higher than adaptive threshold, then center pixel is asterism, described adaptive threshold:
Wherein midvalue is gray-scale value intermediate value, and m (x, y) is local mean value.
2. realize a FPGA device for the lower asterism dividing method of moonlight interference, it is characterized in that comprising: greylevel window generation module, partial gradient module, adaptive threshold module and segmentation result processing module; Wherein:
Described greylevel window generation module, generates the greylevel window data of different size after buffer memory part gradation data, enter partial gradient module and adaptive threshold module respectively; The result of partial gradient module and adaptive threshold module enters segmentation result processing module, if comprise a part for asterism or asterism in partial gradient module determination window, the window center of adaptive threshold module determination simultaneously grey scale pixel value is higher than adaptive threshold, then window center pixel is finally defined as asterism.
3. the FPGA device realizing the lower asterism dividing method of moonlight interference according to claim 2, it is characterized in that: described partial gradient module comprises Sobel wave filter, gradient window generation module, maximum filter, mean filter, Grads threshold computing module and Grads threshold comparison module; Greylevel window data enter Sobel wave filter computed image gradient, and image gradient generates gradient window data and enters maximum filter after gradient window generation module buffer memory, obtains partial gradient; Meanwhile, greylevel window data enter mean filter and calculate local mean value, and local mean value enters Grads threshold computing module and calculates Grads threshold; The partial gradient calculated and Grads threshold enter Grads threshold comparison module, by comparing the part determining whether to comprise asterism or asterism in window.
4. the FPGA device realizing the lower asterism dividing method of moonlight interference according to claim 2, is characterized in that: described adaptive threshold module comprises mean filter, median filter, adaptive threshold computing module and adaptive threshold comparison module; Greylevel window data enter mean filter and median filter calculates local mean value and local intermediate value, the local mean value obtained and local intermediate value enter adaptive threshold computing module, calculate adaptive threshold, the adaptive threshold obtained enters adaptive threshold comparison module, by comparing with window center gray-scale value, determine that whether window center pixel is higher than adaptive threshold.
5. the FPGA device realizing the lower asterism dividing method of moonlight interference according to claim 3, is characterized in that: the maximum filter of described gradient window data is separable, can the filtering of rank of advanced units maximal value, then carries out the filtering of row maximal value.
6. the FPGA device realizing the lower asterism dividing method of moonlight interference according to claim 3, it is characterized in that: described mean filter adopts two one dimension mean filter combinations to realize two-dimentional mean filter, namely when new row enter asterism video in window, calculate new row and, and be added with gray scale summation in video in window, then cut leave video in window the 8th row row and, a sub-addition and a subtraction is only needed to realize mean filter, and without the need to considering image window size, each arranges and needs use twice, once that it is when entering window, once that it is when leaving window, in order to arrange and second time use, adopt FIFO come buffer memory row and.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510288957.5A CN104835177B (en) | 2015-05-29 | 2015-05-29 | A kind of moonlight disturbs lower asterism dividing method and FPGA implement device thereof |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510288957.5A CN104835177B (en) | 2015-05-29 | 2015-05-29 | A kind of moonlight disturbs lower asterism dividing method and FPGA implement device thereof |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104835177A true CN104835177A (en) | 2015-08-12 |
CN104835177B CN104835177B (en) | 2016-05-18 |
Family
ID=53813042
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510288957.5A Active CN104835177B (en) | 2015-05-29 | 2015-05-29 | A kind of moonlight disturbs lower asterism dividing method and FPGA implement device thereof |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104835177B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105374035A (en) * | 2015-11-03 | 2016-03-02 | 北京航空航天大学 | Star sensor star point extraction method under stray light interference |
CN107895384A (en) * | 2017-12-01 | 2018-04-10 | 中国科学院长春光学精密机械与物理研究所 | Target extraction method and device |
CN109685782A (en) * | 2018-12-18 | 2019-04-26 | 北京遥感设备研究所 | A kind of high velocity star point detecting method and system based on FPGA |
CN111402176A (en) * | 2020-04-21 | 2020-07-10 | 中国科学院光电技术研究所 | Method for removing APS star sensor fixed pattern noise in real time on orbit |
CN112634295A (en) * | 2020-12-29 | 2021-04-09 | 中国人民解放军国防科技大学 | Star sensor star point segmentation method based on dual gradient threshold |
CN113514054A (en) * | 2021-06-16 | 2021-10-19 | 北京遥感设备研究所 | Star sensor star point image spot detection method and system |
CN113660413A (en) * | 2021-07-26 | 2021-11-16 | 中国科学院西安光学精密机械研究所 | Automatic exposure method for large-caliber large-view-field camera applied to aircraft |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101976437A (en) * | 2010-09-29 | 2011-02-16 | 中国资源卫星应用中心 | High-resolution remote sensing image variation detection method based on self-adaptive threshold division |
CN102509276A (en) * | 2011-11-25 | 2012-06-20 | 浙江大学 | Weighted constraint-based star atlas segmentation method |
US20140016863A1 (en) * | 2012-07-06 | 2014-01-16 | Samsung Electronics Co., Ltd | Apparatus and method for performing visual search |
-
2015
- 2015-05-29 CN CN201510288957.5A patent/CN104835177B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101976437A (en) * | 2010-09-29 | 2011-02-16 | 中国资源卫星应用中心 | High-resolution remote sensing image variation detection method based on self-adaptive threshold division |
CN102509276A (en) * | 2011-11-25 | 2012-06-20 | 浙江大学 | Weighted constraint-based star atlas segmentation method |
US20140016863A1 (en) * | 2012-07-06 | 2014-01-16 | Samsung Electronics Co., Ltd | Apparatus and method for performing visual search |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105374035A (en) * | 2015-11-03 | 2016-03-02 | 北京航空航天大学 | Star sensor star point extraction method under stray light interference |
CN107895384A (en) * | 2017-12-01 | 2018-04-10 | 中国科学院长春光学精密机械与物理研究所 | Target extraction method and device |
CN109685782A (en) * | 2018-12-18 | 2019-04-26 | 北京遥感设备研究所 | A kind of high velocity star point detecting method and system based on FPGA |
CN111402176A (en) * | 2020-04-21 | 2020-07-10 | 中国科学院光电技术研究所 | Method for removing APS star sensor fixed pattern noise in real time on orbit |
CN111402176B (en) * | 2020-04-21 | 2023-02-14 | 中国科学院光电技术研究所 | Method for removing APS star sensor fixed mode noise in real time on orbit |
CN112634295A (en) * | 2020-12-29 | 2021-04-09 | 中国人民解放军国防科技大学 | Star sensor star point segmentation method based on dual gradient threshold |
CN113514054A (en) * | 2021-06-16 | 2021-10-19 | 北京遥感设备研究所 | Star sensor star point image spot detection method and system |
CN113660413A (en) * | 2021-07-26 | 2021-11-16 | 中国科学院西安光学精密机械研究所 | Automatic exposure method for large-caliber large-view-field camera applied to aircraft |
CN113660413B (en) * | 2021-07-26 | 2022-05-10 | 中国科学院西安光学精密机械研究所 | Automatic exposure method for large-caliber large-view-field camera applied to aircraft |
Also Published As
Publication number | Publication date |
---|---|
CN104835177B (en) | 2016-05-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104835177B (en) | A kind of moonlight disturbs lower asterism dividing method and FPGA implement device thereof | |
CN108230329B (en) | Semantic segmentation method based on multi-scale convolution neural network | |
CN110956126B (en) | Small target detection method combined with super-resolution reconstruction | |
CN109242888B (en) | Infrared and visible light image fusion method combining image significance and non-subsampled contourlet transformation | |
US20190180469A1 (en) | Systems and methods for dynamic facial analysis using a recurrent neural network | |
Kim et al. | High-speed drone detection based on yolo-v8 | |
CN109035172B (en) | Non-local mean ultrasonic image denoising method based on deep learning | |
CN112488999B (en) | Small target detection method, small target detection system, storage medium and terminal | |
CN105046713A (en) | Morphology-based robot star point segmentation method and FPGA realization method | |
CN102944227A (en) | Method for extracting fixed star image coordinates in real time based on field programmable gate array (FPGA) | |
CN111507340A (en) | Target point cloud data extraction method based on three-dimensional point cloud data | |
Chen et al. | Scene segmentation of remotely sensed images with data augmentation using U-net++ | |
CN110969630A (en) | Ore bulk rate detection method based on RDU-net network model | |
Cai et al. | Wavelet-based segmentation on the sphere | |
CN117456376A (en) | Remote sensing satellite image target detection method based on deep learning | |
CN117351371A (en) | Remote sensing image target detection method based on deep learning | |
CN115908409A (en) | Method and device for detecting defects of photovoltaic sheet, computer equipment and medium | |
CN112133100B (en) | Vehicle detection method based on R-CNN | |
CN116543246A (en) | Training method of image denoising model, image denoising method, device and equipment | |
CN114581472A (en) | Image edge detection method and device, electronic equipment and storage medium | |
CN113901903A (en) | Road identification method and device | |
CN113743487A (en) | Enhanced remote sensing image target detection method and system | |
Zhu et al. | YOLO-SDLUWD: YOLOv7-based small target detection network for infrared images in complex backgrounds | |
Zhang et al. | Entropy-Based re-sampling method on SAR class imbalance target detection | |
CN117036985B (en) | Small target detection method and device for video satellite image |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
EXSB | Decision made by sipo to initiate substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |