CN-121982075-A - Satellite-borne strabismus SAR image registration method combining amplitude and phase information

Abstract

The invention relates to a satellite-borne strabismus SAR image registration method combining amplitude and phase information, and belongs to the field of signal processing. The method comprises the following steps of S1, performing coarse registration of a main strabismus SAR image and an auxiliary strabismus SAR image by adopting an amplitude-based correlation function method, so that sub-pixel level registration accuracy can be achieved, and S2, performing fine registration by adopting a method based on frequency spectrum segmentation and differential phase according to the change relation of the phase information of the strabismus SAR image along with the azimuth position. The method solves the problem that the precision of the traditional method based on the amplitude information only does not meet the application requirement under the condition of strabismus SAR images, and has good universality and practicability.

Inventors

• KUANG HUI
• WANG TAO
• LI YI
• WANG YU
• Shan Jingzhe
• JIANG WEN
• MAO YONGFEI
• LIU JIE
• Song Chenying

Assignees

• 中国空间技术研究院

Dates

Publication Date

20260505

Application Date

20251212

Claims (9)

1. 1. The satellite-borne strabismus SAR image registration method combining amplitude and phase information is characterized by comprising the following steps of: s1, performing coarse registration of the main strabismus SAR image and the auxiliary strabismus SAR image by adopting an amplitude-based correlation function method, so that sub-pixel level registration accuracy can be achieved; S2, performing fine registration by adopting a method based on frequency spectrum segmentation and differential phase according to the change relation of the squint SAR image phase information along with the azimuth position.
2. 2. The method for registering an on-board squint SAR image by combining amplitude and phase information according to claim 1, wherein the step S1 specifically comprises: s1.1, selecting pixel points of an area image to be registered from a main SAR image and an auxiliary SAR image, performing image interpolation processing, and calculating a cross-correlation function of the image data of the area to be registered after interpolation of the main SAR image and the auxiliary SAR image; s1.2, searching the position of the maximum value of the cross-correlation function, wherein the corresponding index variable is the position offset of the auxiliary SAR image relative to the main SAR image, and comprises azimuth position offset and distance position offset; s1.3, resampling the auxiliary SAR image by adopting a truncated Sinc interpolation mode according to the azimuth position offset and the distance position offset of the auxiliary SAR image relative to the main SAR image, so that the main SAR image and the auxiliary SAR image realize sub-pixel level coarse registration.
3. 3. The method for registering an on-board squint SAR image in combination with amplitude and phase information according to claim 2, wherein the cross-correlation function formula is as follows: Wherein, the 、 Respectively representing pixel points in the main SAR image and the auxiliary SAR image Is used for the control of the amplitude of (a), Is the coordinates of the pixel points, Representing a pixel point coordinate value range of a region to be registered; 、 The mean and variance of the primary SAR image in the region where registration is required, 、 The mean and variance of the auxiliary SAR image in the region where registration is required, 、 The azimuth offset and the distance-to-position offset of the secondary SAR image relative to the primary SAR image are respectively.
4. 4. The method for registering an on-board squint SAR image by combining amplitude and phase information according to claim 2, wherein the step S2 specifically comprises: S2.1, transforming the main SAR image and the rough registered auxiliary SAR image into an azimuth frequency domain through azimuth Fourier transform, uniformly dividing an azimuth frequency spectrum into an upper part and a lower part according to the frequency, transforming the azimuth frequency spectrums of the upper part and the lower part into an azimuth time domain through azimuth inverse Fourier transform respectively, obtaining the main SAR image and the auxiliary SAR image after dividing the azimuth frequency spectrum respectively, and recording the main SAR image and the auxiliary SAR image as an upper azimuth frequency band main SAR image Main SAR image of lower azimuth frequency band SAR image assisted by upper azimuth frequency band SAR image assisted by lower azimuth frequency band ; S2.2, performing interference processing on the upper position frequency band auxiliary SAR image and the upper position frequency band main SAR image to obtain a main and auxiliary interference phase of an upper position frequency band; S2.3, subtracting the main and auxiliary interference phases of the upper azimuth frequency band from the main and auxiliary interference phases of the lower azimuth frequency band to obtain differential phases; s2.4, calculating registration errors of the main SAR image and the auxiliary SAR image according to the differential phase; S2.5, substituting the registration errors of the main SAR image and the auxiliary SAR image into a primary linear phase compensation function, and multiplying the auxiliary SAR image to the azimuth frequency domain through azimuth Fourier transform and the primary linear phase compensation function to realize azimuth fine registration.
5. 5. The method for registering a strabismus SAR image combining amplitude and phase information as set forth in claim 4, wherein said primary and secondary interference phases of said upper azimuthal band are The method comprises the following steps: Wherein, the 、 The number of azimuth pixels and the number of distance pixels of the image area used for calculating the interference phase are respectively, Represents the range of pixel coordinate values required for the image area used to calculate the interference phase, Assisting SAR image for upper azimuth frequency band Is a conjugate of (c).
6. 6. The method for squint SAR image registration in combination with amplitude and phase information as set forth in claim 5, wherein the primary and secondary interference phases of the upper azimuth band The method comprises the following steps: Wherein, the Assisting SAR image for lower azimuth frequency band Is a conjugate of (c).
7. 7. The method for satellite-borne squint SAR image registration combining amplitude and phase information as set forth in claim 5, wherein the differential phase is The method comprises the following steps: 。
8. 8. the method for satellite borne squint SAR image registration combining amplitude and phase information as set forth in claim 4, wherein the main SAR image and auxiliary SAR image registration errors The method comprises the following steps: Wherein, the For the satellite beam ground scan speed, Is the doppler center frequency of the upper half of the spectrum, Is the doppler center frequency of the lower half of the spectrum.
9. 9. The method for registering an on-board squint SAR image combining amplitude and phase information as set forth in claim 8, wherein the primary linear phase compensation function is: Wherein, the Is the azimuth frequency point.

Description

Satellite-borne strabismus SAR image registration method combining amplitude and phase information Technical Field The invention belongs to the field of signal processing, and relates to a satellite-borne strabismus SAR image registration method combining amplitude and phase information. Background The conventional SAR satellites adopt front side view for two-dimensional imaging, interference imaging and three-dimensional tomography, and play an important role in earth observation. But only small-angle information of an observation area can be acquired by adopting a front-side view mode, multi-angle information cannot be acquired, and along with the development of SAR satellite technology, the satellite-borne SAR is developed towards the multi-angle squint imaging direction in the aspect of application requirements such as three-dimensional deformation monitoring of topography, three-dimensional reconstruction of artificial targets of urban buildings and the like, so that multi-angle squint SAR imaging can be carried out on the same area, and multi-angle information acquisition is realized. When interference imaging or three-dimensional tomography is carried out based on a satellite-borne squint SAR image, the problem of satellite-borne squint SAR image registration needs to be solved firstly, the problems of two aspects are faced, firstly, geometric distortion of the satellite-borne squint SAR image is more serious than that of a front side view image, two images are difficult to register, and secondly, the azimuth registration error of the satellite-borne squint SAR image not only causes mismatching of image amplitude, but also causes phase error, and interference imaging performance, three-dimensional imaging performance and the like are directly influenced. The conventional image registration method based on amplitude matching cannot meet the requirement of satellite-borne squint SAR image registration, so that the research on the registration method of the satellite-borne squint SAR image is required. Disclosure of Invention The invention solves the technical problems of overcoming the defects of the prior art, solving the problem of accurate registration of the satellite-borne strabismus SAR image and laying a technical foundation for multi-angle satellite-borne SAR interference imaging and tomographic three-dimensional imaging. The technical scheme of the invention is that the satellite-borne strabismus SAR image registration method combining amplitude and phase information comprises the following steps: s1, performing coarse registration of the main strabismus SAR image and the auxiliary strabismus SAR image by adopting an amplitude-based correlation function method, so that sub-pixel level registration accuracy can be achieved; S2, performing fine registration by adopting a method based on frequency spectrum segmentation and differential phase according to the change relation of the squint SAR image phase information along with the azimuth position. Preferably, the step S1 specifically includes: s1.1, selecting pixel points of an area image to be registered from a main SAR image and an auxiliary SAR image, performing image interpolation processing, and calculating a cross-correlation function of the image data of the area to be registered after interpolation of the main SAR image and the auxiliary SAR image; s1.2, searching the position of the maximum value of the cross-correlation function, wherein the corresponding index variable is the position offset of the auxiliary SAR image relative to the main SAR image, and comprises azimuth position offset and distance position offset; s1.3, resampling the auxiliary SAR image by adopting a truncated Sinc interpolation mode according to the azimuth position offset and the distance position offset of the auxiliary SAR image relative to the main SAR image, so that the main SAR image and the auxiliary SAR image realize sub-pixel level coarse registration. Preferably, the cross-correlation function is formulated as follows: Wherein, the 、Respectively representing pixel points in the main SAR image and the auxiliary SAR imageIs used for the control of the amplitude of (a),Is the coordinates of the pixel points,Representing a pixel point coordinate value range of a region to be registered;、 The mean and variance of the primary SAR image in the region where registration is required, 、The mean and variance of the auxiliary SAR image in the region where registration is required,、The azimuth offset and the distance-to-position offset of the secondary SAR image relative to the primary SAR image are respectively. Preferably, the step S2 specifically includes: S2.1, transforming the main SAR image and the rough registered auxiliary SAR image into an azimuth frequency domain through azimuth Fourier transform, uniformly dividing an azimuth frequency spectrum into an upper part and a lower part according to the frequency, transforming the azimuth frequency spectrums of t