This paper proposes an absolute phase retrieval method for complex object measurement based on quantized phase-coding and connected region labeling. A specific code sequence is embedded into quantized phase of three coded fringes. Connected regions of different codes are labeled and assigned with 3-digit-codes combining the current period and its neighbors. Wrapped phase, more than 36 periods, can be restored with reference to the code sequence. Experimental results verify the capability of the proposed method to measure multiple isolated objects.

Single-pixel imaging (SPI), which is generally based on computational imaging, has the advantages of a wide bandwidth and the ability to image objects beyond the visual field. However, the major challenge in developing SPI is the large number of illumination patterns that are required. Unlike traditional SPI that relies on random measurement patterns, the SPI method proposed in this letter involves a two-step phase shift that reduces greatly the required number of illumination patterns. Theoretical analysis shows that 6724 illumination patterns are required to reconstruct a 128 × 128-pixel image whose peak signal-to-noise ratio exceeds 30, and these can be projected in 0.3362 s with a digital micromirror device working at full speed. Compared to SPI with a four-step phase shift, half the number of illumination patterns are required. Verification experiments show that the reconstructed images can be obtained even at a sampling ratio of 20%. The proposed SPI with a two-step phase shift is an effective means of requiring fewer illumination patterns and has great potential in dynamic detection.

Vibrational energy harvesters, which are generally based on a single cantilever beam, are mainly effective for excitations perpendicular to the beam. However, typical ambient vibrations are of multifrequency and are multidirectional. To achieve full harvesting of multifrequency vibrational energy in multiple directions, this letter proposes a multidirectional vibrational energy harvester utilizing multimodal vibrations of a double-branched beam that has four natural frequencies below 22 Hz. Comparison experiments in different directions demonstrate an improvement of energy harvesting when this double-branched beam is used. Under horizontal excitation, the maximum open-circuit voltage is increased to 1845%. Under vertical excitation, the average power density is increased over 3800 times. The experimental results indicate that this structural design significantly improves the energy collection efficiency in both horizontal and vertical directions, even with a lighter structural weight.

For conventional camera calibration methods, well-focused images are necessary to detect features accurately. However, this requirement causes practical inconveniences to image acquisition for long- and short-distance photogrammetry. In this study, three active phase-shift circular grating (PCG) arrays are used as calibration patterns. The PCGs’ centers are regarded as feature points that can be accurately extracted by ellipse fitting of $2\pi$-phase points even though patterns are substantially blurred. In the experiments, Gaussian filters are utilized to blur pattern images, and different standard deviations are set for different fuzzy degrees. Pattern images with different defocusing degrees are also captured. The period and number of PCGs and noise are considered. Experimental results indicate that our method is accurate, reliable, and insensitive to image defocusing.

Machine vision techniques, including camera calibration methods, are of great importance for the development of vision-based measurements. However, in multi-camera calibration methods, rapidly constructing accurate geometric relationships among different coordinates is very difficult. Herein, we present a multi-camera calibration method capable of calibrating the intrinsic and extrinsic parameters of four cameras using only a single captured image per camera. Unlike Zhang’s method, which relies on multiple captured images to calibrate the cameras, the method uses a multi-plane stereo target containing multiple fixed planes to which coded patterns are attached. This target greatly reduces the time required for calibration and improves calibration robustness. The proposed method was experimentally compared with traditional camera calibration. The problem affecting the calibration accuracy in single calibration of multiple cameras is that the feature points on the captured images produce occlusion or different degrees of blurring; in the calibration of multiple cameras multiple times, the error accumulation caused by the calibration of two adjacent cameras is solved. This demonstration of a multi-camera calibration method improves camera calibration and provides a new design philosophy, to the best of our knowledge, for machine vision and vision-based measurement.