Fast SAR Autofocus Based on Ensemble Convolutional Extreme Learning Machine

Inaccurate Synthetic Aperture Radar (SAR) navigation information will lead to unknown phase errors in SAR data. Uncompensated phase errors can blur the SAR images. Autofocus is a technique that can automatically estimate phase errors from data. However, existing autofocus algorithms either have poor focusing quality or a slow focusing speed. In this paper, an ensemble learning-based autofocus method is proposed. Convolutional Extreme Learning Machine (CELM) is constructed and utilized to estimate the phase error. However, the performance of a single CELM is poor. To overcome this, a novel, metric-based combination strategy is proposed, combining multiple CELMs to further improve the estimation accuracy. The proposed model is trained with the classical bagging-based ensemble learning method. The training and testing process is non-iterative and fast. Experimental results conducted on real SAR data show that the proposed method has a good trade-off between focusing quality and speed.

Feature Preserving Autofocus Algorithm for Phase Error Correction of SAR Images

Autofocus is an essential technique for airborne synthetic aperture radar (SAR) imaging to correct phase errors mainly due to unexpected motion error. There are several well-known conventional autofocus methods such as phase gradient autofocus (PGA) and minimum entropy (ME). Although these methods are still widely used for various SAR applications, each method has drawbacks such as limited bandwidth of estimation, low convergence rate, huge computation burden, etc. In this paper, feature preserving autofocus (FPA) algorithm is newly proposed. The algorithm is based on the minimization of the cost function containing a regularization term. The algorithm is designed for postprocessing purpose, which is different from the existing regularization-based algorithms such as sparsity-driven autofocus (SDA). This difference makes the proposed method far more straightforward and efficient than those existing algorithms. The experimental results show that the proposed algorithm achieves better performance, convergence, and robustness than the existing postprocessing autofocus algorithms.

High-Speed Focusing and Tracking of Multisized Microbiological Objects

This paper proposes a novel algorithm for high-speed autofocusing and tracking of multisized microbiological objects observed under a transmitted light microscope. Unlike well-known autofocus algorithms found in the literature, the intensity variation of only a small defined region around the border of the microobject is analyzed in the frequency domain to determine the focused position of the object quickly. In the experiment, 20 µm3T3-SWISS cells were used as smallermicroobjects and 97 µm diameter microspheres were used to represent larger microbiological objects. The execution time and accuracy of the proposed algorithmwere assessed and better results were obtained compared to some related autofocusing algorithms. Since its computational cost was low, the algorithm facilitated highspeed autofocusing of both 3T3-SWISS cells and microspheres. The algorithm was also applied to the tracking of moving microobjects by implementing a PD controller. Since visual feedback took only about 1 ms, high-speed tracking was achieved.