Study On ISAR Image Cross-range Scaling Methods

Inverse synthetic aperture radar (ISAR) imaging is an effective means of radar targetrecognition. However, in order to obtain correct information about the target feature, it is desirableto transform image from the conventional range–Doppler domain to the range–cross-rangedomain. Range scaling is easy to complete, but the cross-range scaling factor is difficult todetermine due to the non-cooperation of the targets. Aiming at providing useful scaling methodsin practical applications, three cross-range scaling methods are deeply studied in this thesis.Simulated and real ISAR data are processed to demonstrate the effectiveness of these methods.The scaling performance is analyzed and compared from the perspective of accuracy,computational complexity and applicability.Chapter1is the introduction. The development of the ISAR technique along with the ISARimaging technique is introduced briefly. The research background and main contents of the thesisare presented.In Chapter2, a cross-range scaling method based on the frequency modulation rate(chirp-rate) estimation (FMRE-SM for short) is presented. The modulus of the target effectiverotation vector is estimated by exploiting information carried by the chirp rates of the scattererechoes. A technique based on image segmentation, local polynomial Fourier transform (LPFT),and image contrast (IC) maximisation is used to extract the scattering centres and estimate theirchirp rates. Simulated and real data processing results are provided to confirm the effectiveness ofthe proposed algorithm.In Chapter3, a cross-range scaling method based on the image rotational correlationtechnique (IRC-SM for short) is presented. Two ISAR images are firstly obtained using adjacentdata patches. Then, the rotational velocity is estimated by searching over a range of rotationalvelocities and maximizing the correlation of the two images. Finally, the cross-range scale of theISAR image is obtained and the scaling is performed. Experimental results with both simulatedand real ISAR data are provided to demonstrate the effectiveness of the proposed method.In Chapter4, a cross-range scaling method using the convolution backprojection (CBP)imaging technique (CBP-SM for short) is presented. In this method, the rotational angle betweentwo adjacent pulses is estimated by maximizing the contrast of the resulting CBP images. Thescaling results of the simple point-target ISAR image, simulated ship images and real airplaneISAR images deomonstrate the effectiveness of the scaling method.Chapter5presents a detailed performance comparision of three methods described above from the perspective of accuracy, computational complexity and applicability. CBP-SM has highaccuracy as well as good applicability and is easy to implement. FMRE-SM has a relatively smallcomputational complexity, however with relatively low scaling accuracy. IRC-SM has a relativelyhigh computational complexity, but with good scaling accuracy.Chapter6is the conclusion. The issues that will be studied in the future are pointed out aswell.