PIV Image Background Denoising Algorithm Based on Improved Otsu Thresholding

Particle image velocimetry is an optical measurement technique that uses PIV instruments to measure flow fields. However, due to background light fluctuation, noise interference, voltage fluctuation and other factors, there will be noise interference of different intensities in the background of the collected image. The existing image preprocessing methods cannot handle this situation well. In this paper, a PIV image background interference removal algorithm based on improved neighborhood Otsu processing is proposed. The algorithm proposed in this paper separates the particle image from the background interference through the adaptive neighborhood improved Otsu threshold segmentation method, and uses the common PIV analysis tools PIVLab and paraPIV to analyze the flow field after the interference is removed. The experimental results demonstrated that the proposed algorithm can obviously improve the quality of PIV results in terms of both PSNR and SSIM in the case of the background light interference, and the increasement of average performance is nearly 50 percent when comparing with traditional pre-processing algorithms, which solves the problem of large flow field analysis error caused by poor background light removal effect in the case of irregular grating and other background light interference only using traditional preprocessing.