Dislocation and distortion correction using back-EMF-derived line-data information in compact resonant...

Accurate two-dimensional image reconstruction in compact resonant electromagnetic microelectromechanical system (MEMS) scanning confocal microscopy depends on segmenting a continuously acquired one-dimensional detector stream into image lines and estimating the scan parameters. Mismatch between nominal drive timing and actual fast-axis motion can cause line dislocation and geometric distortion, whereas additional sensing hardware, external synchronization, or image-domain estimation may increase instrument size and implementation complexity. We present a method that obtains the required line-data information directly from the fast-axis back electromotive force (back-EMF). Under constant-current H-bridge drive, the fast-axis coil-terminal voltage was acquired by pseudo-differential sampling and processed to extract the back-EMF waveform. From the extracted waveform, a field-programmable gate array (FPGA) generated center-aligned line data, the line-scan period, and the optical scan amplitude in real time. The host computer used these outputs to perform scan-parameter-based two-dimensional image reconstruction. For a standard square-grid target, the line-offset standard deviation decreased from 4.397 to 0.618 pixels, a reduction of approximately 86%. The mean grid-cell aspect ratio improved from 1.1282 to 1.0074, closer to the ideal value of unity, while its mean absolute error (MAE) relative to unity decreased from 0.1282 to 0.0082, a reduction of approximately 93.6%. Imaging of a plant leaf showed improved interline continuity and local morphological consistency in complex cellular structures. The current system displayed 1000×1000-pixel images in real time at 30 fps, demonstrating the feasibility of this route for acquiring back-EMF-derived line-data information and applying it to image correction in a compact high-frame-rate imaging configuration.