CN112346057B - Two-dimensional interpolation module in image processing process - Google Patents
Two-dimensional interpolation module in image processing process Download PDFInfo
- Publication number
- CN112346057B CN112346057B CN201910719523.4A CN201910719523A CN112346057B CN 112346057 B CN112346057 B CN 112346057B CN 201910719523 A CN201910719523 A CN 201910719523A CN 112346057 B CN112346057 B CN 112346057B
- Authority
- CN
- China
- Prior art keywords
- interpolation
- data
- module
- distance
- azimuth
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a two-dimensional interpolation module in an image processing process, which comprises a multi-path data distribution module, an interpolation module and a multi-path data integration module, wherein the multi-path data distribution module comprises a plurality of data distribution modules and a plurality of interpolation modules: the multi-path data distribution module outputs data in a time-sharing way, and streams the data into the parallel interpolation processing module; interpolation module: the system comprises a distance interpolation module and a direction interpolation module; respectively calculating the interpolation positions of the distance direction and the azimuth direction and interpolating; and the multi-path data integration module integrates interpolation results output by the distance and azimuth interpolation processing into one path respectively for output. The real-time performance of the module is strong, and the image data processing speed is remarkably improved.
Description
Technical field:
the invention can be applied to the field of image processing, and particularly relates to imaging signal processing of a synthetic aperture radar in a beam-focusing mode.
Background
In the current focusing imaging of synthetic aperture radars (Synthetic Aperture Radar, SAR) using a beam-focusing mode, a polar format algorithm (Polar Format Algorithm, PFA) is generally used to perform preliminary processing on received radar echo data, and then a two-dimensional fast fourier transform (FAST Fourier Transform, FFT) is performed to realize focusing imaging of a target. However, the actual radar echo data are mapped to the spatial frequency domain and are arranged in a polar coordinate format, and the requirement that two-dimensional FFT needs to be sampled orthogonally and uniformly in the spatial frequency domain cannot be met, so that the polar coordinate format data need to be converted into rectangular format data, namely, the distance direction and the azimuth direction interpolation need to be carried out first.
The traditional interpolation is single-path serial processing, consumes a great amount of CPU running time and has poor real-time performance, which is unacceptable for a radar imaging system with extremely high real-time requirements, so that the interpolation process becomes one of the bottlenecks for limiting the radar signal processing speed.
Disclosure of Invention
Aiming at the technical problem of poor instantaneity of an interpolation process, the invention provides a two-dimensional interpolation module in an image processing process, which is based on a Field programmable gate array (Field-Programmable Gate Array, FPGA) device, skillfully provides a distance-oriented and azimuth-oriented parallel interpolation operation structure suitable for a large-data-volume occasion, and accelerates the interpolation speed and saves storage resources by a linear derivation method. The real-time performance of the module is strong, and the image data processing speed is remarkably improved.
The invention discloses a two-dimensional interpolation module in an image processing process, which is characterized by comprising a multi-path data distribution module, an interpolation module and a multi-path data integration module:
the multi-path data distribution module comprises a multi-path distance direction data distribution module and a multi-path azimuth direction data distribution module;
the multipath distance direction data distribution module: dividing the preprocessed input data into multiple channels, outputting the multiple channels in a time-sharing way, and sending the running water into a parallel distance interpolation processing module;
the multi-path azimuth data distribution module comprises: dividing the data subjected to distance data interpolation into multiple channels, outputting the multiple channels in a time-sharing way, and sending the multiple channels into parallel azimuth interpolation processing modules;
interpolation module: the system comprises a distance interpolation module and a direction interpolation module; respectively calculating the interpolation positions of the distance direction and the azimuth direction and interpolating;
and the multi-path data integration module integrates interpolation results output by the distance and azimuth interpolation processing into one path respectively for output.
Further, the distance direction interpolation module calculates a distance direction interpolation position by adopting a derivation method.
Further, the distance interpolation module is:
interpolation position X is
Will interpolate position X n Decomposition into integer parts M n And fractional part L n Then
Outputting the interpolated signal Dout with the expression of
Dout(n+1)＝[Din(M n +1)-Din(M n )]·L n +Din(M n )
n is the number of data blocks; din is the input radar echo signal; initial value K of distance direction data before and after interpolation y_in0 、K y_out0 And delta value delta K y_in 、ΔK y_out The distance-interpolated signal Dout is output.
Further, the azimuth interpolation module calculates azimuth interpolation positions by adopting a search method.
Further, the azimuth interpolation module is:
will K x_out 4096 data are sequentially combined with K from big to small or from small to big x_in The data in (a) are compared to obtain nth data K x_out (n) at K x_in Positions in the sequence, which are the interpolated position integer part M n And get a position at K x_out (n) data K on left side x_in (M n ) And data K on the right side x_in (M n +1)；
After obtaining the interpolation position M and the corresponding data, if the interpolation position decimal part L needs to be calculated
Estimating according to the interpolation position M and the distance-direction interpolation data Dout to obtain an interpolated signal Dout', wherein the expression is as follows
Dout'(n+1)＝[Dout(M n +1)-Dout(M n )]·L n +Dout(M n )。
Further, the number of the distance distribution channels is the same as that of the distance interpolation modules; the number of the azimuth distribution channels is the same as that of the azimuth interpolation modules.
Further, the module operates on a field programmable gate array.
The beneficial effects are that:
(1) Interpolation calculation is realized by selecting a reasonable parallel pipeline structure, so that the interpolation processing speed is greatly improved, the calculation speed of linear interpolation and performance indexes such as hardware resource occupation are optimized, and the method can be widely applied to radar imaging processing.
(2) For the data with the uniformly distributed coordinate directions, an interpolation position derivation method is used for obtaining interpolation positions, all coordinate data in the distance direction and the azimuth direction are not required to be stored, only the derivation is required according to the initial value and the increment value, storage resources can be effectively saved, and the interpolation calculation efficiency is greatly improved. For data with non-uniform distribution in the coordinate direction, an interpolation position searching method is used for obtaining interpolation positions, the comparison of the sizes of all data is not needed, and only the data before and after the interpolation positions are found, so that the efficiency of interpolation calculation can be improved.
(3) The advantages of FPGA full parallel pipeline processing, low power consumption, abundant triggers and I/O pins and the like are brought into play, and the quick operation of linear interpolation is completed, so that the performance indexes such as the speed, the precision and the hardware cost of the linear interpolation are optimal.
The method has the advantages of higher processing speed, capability of solving the problem that software such as a digital signal processor and the like cannot meet the real-time requirement of the radar imaging system, and greatly improving the speed of interpolation operation so as to meet the requirement of the radar imaging system on higher real-time requirement of linear interpolation.
Drawings
FIG. 1 is a schematic flow chart of the present invention
FIG. 2 is a schematic diagram illustrating a distance-oriented data distribution module timing sequence according to an embodiment
FIG. 3 is a schematic diagram illustrating a timing sequence of a direction data distribution module according to an embodiment
FIG. 4 is a diagram showing a search process of direction interpolation positions in an embodiment
FIG. 5 is a schematic diagram illustrating a timing sequence of a distance data integration module according to an embodiment
FIG. 6 is a timing diagram of an azimuthal data integration module according to an embodiment
Detailed Description
Before two-dimensional interpolation calculation is performed on radar echo data, distance matched filtering and motion compensation on the center of a scene are performed on the received two-dimensional echo signals, and a pre-interpolation signal can be obtained, which is a preprocessing process. The preprocessing process is a common means in the prior art, and is based on the assumption of plane wave front, the distance direction K before interpolation is obtained y_in From the interpolated distance direction K y_out Direction of azimuth K before interpolation x_in And the direction K after interpolation x_out And sending the data as input data to an interpolation processing module. The input data contains two coordinate direction interpolation front and rear data (K y_in ，K y_out ，K x_in ，K x_out ) And radar echo data (Din) that is distance matched filtered and motion compensated for the center of the scene.
Since a large amount of time is consumed in interpolation processing, and the resource waste is caused by excessively long waiting time after data input, the invention uses a multi-path parallel structure to process the data. The system specifically comprises a multipath data distribution module, an interpolation module and a multipath data integration module.
1. Multi-path data distribution module
The multipath data distribution module comprises a multipath distance direction data distribution module and a multipath azimuth direction data distribution module.
1. The multipath distance direction data distribution module;
dividing the preprocessed input data into channels with the same number as the distance interpolation modules, outputting the input data in a time-sharing way through different channels, and sending the running water into the parallel distance interpolation processing modules.
The distance direction and the azimuth direction data block size selected in this embodiment are 4096 points.
The distance direction position is calculated by adopting a deduction method, so that the calculated amount is small, the input data is divided into two paths by the distance direction data distribution module and is output in a time-sharing way, and the input data is sent into the distance direction interpolation processing module, wherein the time sequence diagram is shown in fig. 2. From the figure, it can be seen that channel 0 and channel 1 respectively send input data at different moments, which lays a foundation for the subsequent parallel distance interpolation processing of two paths.
2. A multi-path azimuth data distribution module;
the data after the distance data interpolation is divided into channels with the same number as the azimuth interpolation modules, the data is output in a time-sharing way through different channels, and the running water is sent into the parallel azimuth interpolation processing modules.
And determining the azimuth interpolation position by adopting a search method. Compared with the deduction method, the search method has larger calculation amount and needs longer time for calculation, in this embodiment, the azimuth data distribution module divides the input data into five paths for time-sharing output, and sends the five paths of time-sharing output data to the azimuth interpolation processing module, and the timing diagram is shown in fig. 3. From the figure, it can be seen that the channels 0, 1, 2, 3 and 4 respectively transmit the input data at different moments, which lays a foundation for the subsequent five-path parallel azimuth interpolation processing.
The data amount and the calculation efficiency in different embodiments are different, and the number of paths distributed by the data distribution module can be adjusted to minimize the processing time.
2. Interpolation module
The interpolation module comprises a distance interpolation module and an azimuth interpolation module. Both include computing interpolation locations and interpolation. Calculating the interpolation position needs to obtain the coordinate directions K before and after interpolation in 、K out The size relationship of the data, record K out Each data corresponds to K in As interpolation positions.
1. The function of the distance interpolation module is to calculate the distance interpolation position and interpolate, the data of the sector distribution in the space domain is converted into the data of the trapezoid distribution, and the input data is the radar echo signal Din and the initial value K of the distance direction data before and after interpolation y_in0 、K y_out0 And delta value delta K y_in 、ΔK y_out The distance-interpolated signal Dout is output.
(1) Deriving distance interpolation position
Due to two distance coordinate data K before and after distance interpolation y_in And K y_out The distribution is uniform, and the initial value and the increment value are known, so that two coordinates do not need to be storedAll data of the directions, while the interpolation position is deduced according to the linear deducing module, a great deal of processing time can be saved. The calculation of the interpolation position X can be simplified to
Will interpolate position X n Is decomposed into an integer part and a fractional part
n is the number of data blocks.
(2) Interpolation
Estimating according to the interpolation position and the input radar echo data Din to obtain an output interpolated signal Dout, wherein the expression is
Dout(n+1)＝[Din(M n +1)-Din(M n )]·L n +Din(M n )
The distance interpolation module is 2 groups, and comprises two parts of deriving distance interpolation positions and interpolation. The calculation efficiency of the FPGA for carrying out the deduction interpolation on the data with the size of 4096 points is 0.5, namely the clock period required by inputting the data sequence is T, and the interpolation calculation is started after the sequence input is finished, and the interpolation is completed by 2T clock periods. The two distance interpolation processing modules perform linear interpolation processing in a ping-pong manner, so that the full-flow water can be approximately achieved, the calculation efficiency is approximately 1, the processing speed is increased, and the processing time is reduced.
2. The azimuth interpolation module is used for searching azimuth interpolation positions and interpolating, converting data which are obtained through distance interpolation and are in trapezoid distribution in a space domain into data in rectangular distribution, inputting the data into signals which are subjected to distance interpolation, and outputting the signals which are subjected to distance interpolation and azimuth interpolation. Wherein, azimuth data K x_in Is non-uniform, and the interpolated azimuth data K x_out The coordinate interval is uniform, so that the azimuth interpolation is a non-uniform to uniform resampling process, and cannot be usedThe deduction method needs to compare the magnitude of each direction data to be interpolated with the magnitude of the original direction data, and searches to obtain the interpolation position. The specific searching process is shown in fig. 4, and comprises the following steps:
(1) The FPGA respectively carries out K x_in And K x_out Writing K into FIFO memory x_out Inner n data are sequentially combined with K x_in The data of (a) are compared in order from large to small or from small to large (based on the radar data order) to obtain K x_out (n) at K x_in Positions in the sequence, which are the interpolated position integer part M n And get a position at K x_out (n) data K on left side x_in (M n ) And data K on the right side x_in (M n +1)。
After obtaining the interpolation position M and the corresponding data, if the interpolation position decimal part L needs to be calculated
(2) Estimating the interpolation processed data Dout according to the interpolation position and the input distance to obtain an output interpolated signal Dout', wherein the expression is as follows
Dout'(n+1)＝[Dout(M n +1)-Dout(M n )]·L n +Dout(M n )
The comparison process is shown in fig. 4, and specifically includes: in K x_out (n) is the reference K x_in Is compared with each of the data of (a). K (K) x_in Counting by n' (0-4095), K x_out Counting by n (0-4095), when K x_in (n')＜K x_out (n) at the time of K x_in A read operation of (n' +1); when K is x_in (n')＞K x_out (n) at the time, K is applied to the FIFO x_out A read operation of (n+1); while when K is x_in (n')＝K x_out (n) at the time of K to the FIFO storing the data and the data x_in (n' +1) and K x_out (n+1) read operations. When reading the FIFO, the data are sequentially read out according to the mode of input first and output first, and then the next round of comparison is carried out.
In this embodiment, the calculation efficiency of the FPGA performing search interpolation on the data size 4096 points is 0.2, that is, 5T clock cycles are required to complete interpolation, so that five interpolation modules need to be instantiated to perform parallel interpolation processing to save the time of data processing, and the interpolation processing is performed between the five interpolation modules in a ping-pong manner, so that the calculation efficiency is approximately 1, the processing speed is increased, and the processing time is reduced.
3. A multi-path data integration module;
and the multi-path data integration module integrates interpolation results output by the distance and azimuth interpolation processing into one path respectively for output.
The distance-wise data integration timing diagram is shown in fig. 5. As can be seen from the figure, the data outputted from the two interpolation processing modules are sent to the data integration module in a time-sharing manner.
The azimuthal data integration timing diagram is shown in FIG. 6. As can be seen from the figure, the data outputted by the five interpolation processing modules are sent to the data integration module in a time sharing manner.
After the interpolation processing is completed, radar echo data which are rectangular in distribution in a space domain can be obtained, and a foundation is laid for subsequent PFA processing.
The foregoing detailed description is provided for the purpose of illustration and description only, and is not intended to limit the scope of the claims. Any simple modification or replacement of the technical scheme of the invention by a person skilled in the art falls within the protection scope of the invention.
Claims (5)
1. The two-dimensional interpolation module is used for interpolating radar echo data in the image processing process and is characterized by comprising a multipath data distribution module, an interpolation module and a multipath data integration module:
the multi-path data distribution module comprises a multi-path distance direction data distribution module and a multi-path azimuth direction data distribution module;
the multipath distance direction data distribution module: dividing the preprocessed input data into multiple channels, outputting the multiple channels in a time-sharing way, and sending the running water into a parallel distance interpolation processing module;
the multi-path azimuth data distribution module comprises: dividing the data subjected to distance data interpolation into multiple channels, outputting the multiple channels in a time-sharing way, and sending the multiple channels into parallel azimuth interpolation processing modules;
interpolation module: the system comprises a distance interpolation module and a direction interpolation module; respectively calculating the interpolation positions of the distance direction and the azimuth direction and interpolating;
the multi-path data integration module integrates interpolation results output by interpolation processing of the distance and azimuth directions into one path for output;
the distance direction interpolation module calculates a distance direction interpolation position by adopting a deduction method, and the distance direction interpolation module is as follows:
interpolation position X is
Will interpolate position X n Decomposition into integer parts M n And fractional part L n Then
Outputting the interpolated signal Dout with the expression of
Dout(n+1)＝[Din(M n +1)-Din(M n )]·L n +Din(M n )
n is the number of data blocks; din is the input radar echo signal; initial value K of distance direction data before and after interpolation y_in0 、K y_out0 And delta value delta K y_in 、ΔK y_out The distance-interpolated signal Dout is output.
2. The two-dimensional interpolation module in an image processing process according to claim 1, wherein the azimuth interpolation module calculates azimuth interpolation positions by using a search method.
3. The two-dimensional interpolation module in an image processing process according to claim 2, wherein the azimuth interpolation module is:
will K x_out 4096 data are sequentially combined with K from big to small or from small to big x_in The data in (a) are compared to obtain nth data K x_out (n) at K x_in Positions in the sequence, which are the interpolated position integer part M n And get a position at K x_out (n) data K on left side x_in (M n ) And data K on the right side x_in (M n +1)；
After obtaining the interpolation position M and the corresponding data, if the interpolation position decimal part L needs to be calculated
Estimating according to the interpolation position M and the distance-direction interpolation data Dout to obtain an interpolated signal Dout', wherein the expression is as follows
Dout'(n+1)＝[Dout(M n +1)-Dout(M n )]·L n +Dout(M n )
K x_in Is azimuth data; k (K) x_out The direction data after interpolation; k (K) x_out (n) is the interpolated nth azimuth data; m is M n For the nth data K x_out (n) at K x_in Positions in the sequence, which are integer parts of the interpolated positions, L n A decimal part for the interpolation position; k (K) x_in (M n ) Is positioned at K x_out (n) data on the left side; k (K) x_in (M n +1) is located at K x_out (n) data on the right side.
4. The two-dimensional interpolation module in an image processing process according to claim 1, wherein the number of the distance distribution channels is the same as the number of the distance interpolation modules; the number of the azimuth distribution channels is the same as that of the azimuth interpolation modules.
5. A two-dimensional interpolation module in an image processing process according to claim 1, wherein said module operates on the basis of a field programmable gate array.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910719523.4A CN112346057B (en) | 2019-08-06 | 2019-08-06 | Two-dimensional interpolation module in image processing process |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910719523.4A CN112346057B (en) | 2019-08-06 | 2019-08-06 | Two-dimensional interpolation module in image processing process |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112346057A CN112346057A (en) | 2021-02-09 |
| CN112346057B true CN112346057B (en) | 2023-07-14 |
Family
ID=74366309
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910719523.4A Active CN112346057B (en) | 2019-08-06 | 2019-08-06 | Two-dimensional interpolation module in image processing process |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112346057B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105223572A (en) * | 2015-09-14 | 2016-01-06 | 北京航空航天大学 | A kind of positive forward sight Bistatic SAR image processing method based on PFA algorithm |
| CN108120980A (en) * | 2017-12-13 | 2018-06-05 | 南京航空航天大学 | A kind of implementation method of the FPGA of satellite-borne SAR multi-modal imaging signal processing algorithm |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105866774A (en) * | 2016-03-23 | 2016-08-17 | 南京航空航天大学 | FPGA implementation method for polar coordinate format imaging algorithm of chirp signal |
| CN109932718B (en) * | 2019-03-11 | 2022-11-04 | 南京航空航天大学 | Multi-rotor unmanned aerial vehicle-mounted circular track all-round-looking SAR (synthetic aperture radar) imaging method |
-
2019
- 2019-08-06 CN CN201910719523.4A patent/CN112346057B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105223572A (en) * | 2015-09-14 | 2016-01-06 | 北京航空航天大学 | A kind of positive forward sight Bistatic SAR image processing method based on PFA algorithm |
| CN108120980A (en) * | 2017-12-13 | 2018-06-05 | 南京航空航天大学 | A kind of implementation method of the FPGA of satellite-borne SAR multi-modal imaging signal processing algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112346057A (en) | 2021-02-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102680945B (en) | Doppler modulation frequency estimation method based on field programmable gate array (FPGA) | |
| CN101373181B (en) | Method and apparatus for calculating point-to-point trace-changing coefficient in real time | |
| CN106896360A (en) | A kind of FPGA implementation method of SAR signal processing algorithms | |
| CN109581275B (en) | Two-dimensional underwater DOA estimation method and device based on non-circular signal and three-dimensional orthogonal array | |
| CN109444901A (en) | A kind of multiple submatrixes SAS sub-aperture image method under isomerous environment | |
| CN105137428A (en) | Dechirp signal polar format imaging algorithm FPGA (Field Programmable Gate Array) realization method | |
| CN103257341A (en) | Fast autofocus algorithm implementation method based on FPGA | |
| CN101403791B (en) | Fast real-time space spectrum estimation ultra-resolution direction-finding device and method thereof | |
| CN113341377B (en) | Radar baseband module and radar system | |
| CN103776907A (en) | Ultrasonic phased array received signal fine delaying method based on sinc interpolation | |
| CN112346057B (en) | Two-dimensional interpolation module in image processing process | |
| CN108614234B (en) | Direction-of-arrival estimation method based on multi-sampling snapshot co-prime array received signal fast Fourier inverse transformation | |
| CN101426134A (en) | Hardware device and method for video encoding and decoding | |
| CN113156435B (en) | Missile-borne SAR front side view time domain imaging method based on embedded GPU | |
| CN116430382A (en) | Real-time imaging processing system and method based on embedded GPU and FPGA isomerism | |
| CN102654574B (en) | Doppler centroid estimation method based on field programmable gate array (FPGA) | |
| CN109633640A (en) | A kind of ISAR Processing Algorithm based on to marine origin picture | |
| Liu et al. | FPGA implementation of SAR imaging processing system | |
| CN114706078A (en) | Terahertz synthetic aperture radar imaging system and method based on digital signal processing | |
| CN111865385B (en) | Two-dimensional planar array digital beam forming method based on FPGA | |
| CN111025293A (en) | Efficient FPGA real-time imaging system applied to small satellite SAR | |
| Chaudhari et al. | An optimized approach to pipelined architecture for fast 2D normalized cross-correlation | |
| CN111221496B (en) | Method for realizing floating point data accumulation by using FPGA | |
| CN110058247A (en) | A kind of method of synthetic aperture sonar real time imagery | |
| Di et al. | FPGA-Based Multi-core Reconfigurable System for SAR Imaging |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |