Existing methods to eliminate the laser speckle noise in quantitative phase imaging always suffer from the loss of detailed phase information and the resolution reduction in the reproduced image. To overcome these problems, this paper proposes a speckle noise suppression method based on empirical mode decomposition. Our proposed method requires only one image without additional equipment and avoids the complicated process of searching the optimal processing parameters. In this method, we use empirical mode decomposition to highlight the high frequency information of the interference image and use the Canny operator to perform edge detection, so the diffusion denoising process is guided by high-precision detection results to achieve better results. To validate the performance of our proposed method, the phase maps processed by our proposed method are compared with the phase maps processed by the improved anisotropic diffusion equation method with edge detection, the mean filter method and the median filter method. The experimental results show that the method proposed in this paper not only has a better denoising effect but also preserves more details and achieves higher phase reconstruction accuracy.

Imaging the interaction between the laser pulse and photoresist mixture on the ultrafast time scale can track the path of the light pulse and reveal the procedure of the microstructure machining. However, most existing imaging technologies suffer from problems such as requiring multiple repeated shots or a limited time resolution. To overcome these problems, we propose to capture the motion of laser pulses in a photoresist mixture by using compressed ultrafast photography (CUP). In this method, we can recover the motion process of non-repeatable events with a single shot at the time-resolution of about $1.54\times {10}^{11}$ fps, where the depth of the imaging sequence reaches hundreds of frames. To verify the effectiveness of the proposed method, we estimate the speed of the laser pulse in a photoresist mixture and evaluate the similarity between the image captured by a streak camera and our reconstructed ultrafast sequence; the results validate the reliability of our proposed method.