Sub-image back projection parallel imaging method based on GPU

Abstract

The invention discloses a sub-image back projection parallel imaging method based on a GPU. The method comprises the following steps: adaptive blocking processing of spaceborne synthetic aperture radar echo data is carried out; gPU parallel backward projection processing is carried out; performing backward projection calculation and data block asynchronous transmission processing; and sub-image generation. According to the method, the backward projection processing algorithm is accelerated by utilizing the multi-thread high-concurrency characteristic of the GPU, the running performance of a program is improved, the practicability is high, the method can be used for high-orbit and high-resolution spaceborne synthetic aperture radar imaging, the defects that the processing time is long due to the fact that the calculation complexity of a traditional synthetic aperture radar backward projection algorithm is high are overcome, and the processing efficiency is improved. Wide market application prospects are realized.

Description

Technical Field

[0001] The present invention belongs to the field of synthetic aperture radar imaging, and mainly relates to a sub-image back projection (BP) parallel imaging method based on a graphics processing unit (GPU). The method utilizes the high-concurrency threads of the GPU to adaptively process spaceborne synthetic aperture radar echo data in blocks and perform sub-image back projection parallel processing, so as to realize SAR imaging efficiently and accurately. Background Art

[0002] The backprojection imaging algorithm is suitable for complex observation geometry and can achieve high-precision spaceborne synthetic aperture radar (SAR) imaging without approximation. As the width / resolution ratio of spaceborne SAR increases, the impact of surface curvature cannot be ignored. Track curvature and surface curvature make the spatial variation of the signal complex. For such problems, the backprojection imaging algorithm can be used to obtain high-longitude focused SAR images.

[0003] In order to solve the problem of large computational complexity of BP algorithm, many scholars have improved BP imaging algorithm based on hierarchical technology, decomposing SAR images into multiple sub-aperture images, also called sub-images, such as Fast Factorized BackProjection (FFBP) algorithm to realize the synthesis of sub-aperture images in polar coordinate system, and Cartesian Factorized Back Projection (CFBP) algorithm to realize the synthesis of sub-aperture images in rectangular coordinate system, which can avoid the problem of image geometric deformation. However, for satellite-borne high-orbit SAR data, these methods still have the problems of large computational complexity and serious time consumption when calculating each level of sub-aperture images, which limits its practical application. Therefore, it is necessary to develop a GPU-based sub-aperture image back-projection parallel imaging method for satellite-borne high-orbit SAR imaging. This method uses the high concurrent threads of GPU to efficiently and accurately realize high-resolution SAR imaging, which has important engineering value and application prospects. Summary of the invention

[0004] In order to achieve the above technical task, the present invention adopts the following technical solution to achieve it: The present invention provides a GPU-based sub-aperture image back-projection parallel imaging method, the method comprising the following steps:

[0005] (1) Adaptive block processing of spaceborne synthetic aperture radar (SAR) echo data; (2) GPU parallel backprojection processing; (3) Backprojection calculation and data block asynchronous transmission processing; (4) Sub-aperture image generation.

[0006] In the step (1), the amount of spaceborne synthetic aperture radar (SAR) echo data is large and cannot be transferred to the GPU memory at one time. First, the data of each position in the sub-aperture image grid of size K is set to zero and transferred to the GPU memory, and then the real-time remaining memory is obtained by using the graphics card driver interface function, and the adaptive segmentation of the echo data is realized based on this.

[0007] This technology maximizes the utilization of video memory, thereby ensuring that the data throughput of subsequent concurrent processing is maximized. At the same time, this technology has good versatility and portability, and is not limited by the type of graphics card and the size of echo data.

[0008] In the step (2), the block echo data is transferred to the GPU video memory, T GPU threads are started, each thread reads a data in the echo data block, and calculates the echo signal strength of a point on the sub-aperture grid through the back projection algorithm, and adds this echo signal strength to the existing value at that position.

[0009] In the step (3), when the number of data blocks N>1, while executing the back projection processing in step (2), since the GPU has independent data transmission hardware, the next block of data can be transferred from the memory to the video memory at the same time. Since the back projection calculation and the data block transmission are asynchronous at this time, this technology is called asynchronous transmission.

[0010] In the step (4), steps (2) and (3) are repeated until all echo data have completed the back-projection processing, at which time the data in the sub-aperture image grid is the final sub-aperture image data.

[0011] The beneficial effects of the invention are as follows: the invention provides an accurate and effective high-resolution synthetic aperture radar imaging calculation method, which is novel and highly practical, can be used for high-orbit, high-resolution spaceborne synthetic aperture radar imaging, overcomes the defects of high computational complexity of the traditional synthetic aperture radar backprojection algorithm, which leads to lengthy processing time, and has broad market application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 is the back-projected sub-aperture image geometry model

[0013] Figure 2 This is the parallel calculation flow chart of the back-projected sub-aperture image. DETAILED DESCRIPTION

[0014] The specific contents of the present invention are further described in detail below in conjunction with the accompanying drawings. Figure 1 and Figure 2 The specific implementation steps of the present invention are as follows:

[0015] Step 1: Adaptive block processing of spaceborne synthetic aperture radar (SAR) echo data. First, arrange the imaging grid, set the data of each position in the sub-aperture image grid of size K to zero, and transfer it to the GPU memory. Then use the graphics driver interface function to obtain the real-time remaining memory size R. Assuming that the total size of the echo data is E, the number of echo data blocks N can be calculated by the following formula:

[0016] N=(E×2+R-1) / R

[0017] The size M1 of the first N-1 blocks of echo data is:

[0018] M1=(E×2+N-1) / N

[0019] The size M2 of the Nth block of echo data is:

[0020] M2=E×2-(N-1)×M1.

[0021] Step 2: GPU parallel back-projection processing, transferring the block echo data of size M1 to the GPU memory. If it is the last data block, the block echo data of size M2 is transferred to the GPU memory.

[0022] The number of GPU threads opened is T.

[0023] T=K×M1

[0024] Each thread reads a data in the data block and calculates the azimuth slow time t corresponding to the data. a The distance R from the radar phase center to the subaperture image grid point p is p (t a )

[0025]

[0026] Among them, (x s ,y s ,z s ) is t a The position coordinates of the radar at the moment in the earth's fixed coordinate system, (x p ,y p ,z p ) is t a The position coordinates of the grid point p on the sub-aperture image at the moment in the earth-fixed coordinate system. Then calculate the echo signal strength at point p

[0027]

[0028] Among them, t r is the distance-to-speed time, t a is the azimuth slow time, K c =4πfc / c Radar carrier center wave number, f c is the radar carrier frequency, π is the circumference of a circle, c is the speed of light, j is the imaginary unit, B is the transmitter bandwidth, and the sinc function is defined as follows:

[0029]

[0030] Step 3: When the number of data blocks N>1, while executing the back-projection processing in step 2, since the GPU has independent data transmission hardware, the next raw echo data block can be transferred from the memory to the video memory at the same time.

[0031] Step 4, repeat steps 2 and 3 until all echo data have completed the backprojection processing, at which point the data in the sub-aperture image grid is the final sub-aperture image data.

[0032] The present invention is not limited to the above-mentioned specific implementation modes. Various changes made by ordinary technicians in this field based on the above-mentioned concepts without creative work are all within the protection scope of the present invention.

Claims

1. A GPU-based sub-image back-projection parallel imaging method, characterized in that: The following steps are involved: (1) Adaptive block processing of spaceborne synthetic aperture radar echo data; (2) GPU parallel back-projection processing; (3) Back-projection calculation and data block asynchronous transmission processing; (4) Sub-image generation.

2. The GPU-based sub-image back-projection parallel imaging method according to claim 1, characterized in that: In step (1), the data at each position in the sub-image grid of size K is first set to zero and transferred to the GPU memory. Then, the real-time remaining memory size R is obtained using the graphics driver interface function. Assuming that the total size of the echo data is E, the number of echo data blocks N can be calculated by the following formula: N=(E×2+R-1) / R The size M1 of the first N-1 blocks of echo data is: M1=(E×2+N-1) / N The size M2 of the Nth block of echo data is: M2=E×2-(N-1)×M1.

3. The GPU-based sub-image back-projection parallel imaging method according to claim 1, characterized in that: In step (2), the block echo data of size M1 is transferred to the GPU memory. If it is the last data block, the block echo data of size M2 is transferred to the GPU memory. The number of GPU threads opened is: T=K×M1 Each thread reads a data in the echo data block, and calculates the echo signal strength of a point on the sub-aperture grid through the back-projection algorithm, and adds this echo signal strength to the existing value at that position.

4. The GPU-based sub-image back-projection parallel imaging method according to claim 1, characterized in that: In step (3), when the number of data blocks N>1, while executing the back-projection processing in step (2), the GPU has independent data transmission hardware and simultaneously transfers the next raw echo data block from the memory to the video memory. At this time, the back-projection calculation and the raw echo data block are asynchronously transmitted.

5. The GPU-based sub-image back-projection parallel imaging method according to claim 1, characterized in that: In step (4), the sub-image is generated by repeating steps (2) and (3) until all echo data have completed the back-projection processing. At this time, the data in the sub-image grid is the final sub-image data.

Images