Adaptive local threshold segmentation for Fourier spatial filtering in automatic analysis of digital speckle interferogram

Automated extraction of the +1 term spectrum can greatly benefit the real-time measurement of dynamic deformation in the digital speckle pattern interferometry (DSPI). However, most spatial filtering methods are not satisfactory for automatic analysis because they need manual intervention or are unable to accurately extract multiple +1 term spectrums in one frequency distribution image. We propose an adaptive local threshold segmentation method for the Fourier spatial filtering of a three-dimensional DSPI system. This method first uses global thresholding to coarsely recognize the spectral regions and then adopts local thresholding to accurately determine the spatial filtering windows for each spectrum. Experimental result shows that the proposed adaptive local threshold segmentation can accurately extract multiple desired +1 term spectrums even if the intensities of different spectral regions differ greatly. In addition, experiments of the comparison between the automatic and manual spatial filtering and automatic analysis of dynamic deformation demonstrate that the proposed method can be used for the real-time measurement.