Adaptive enhancement of micro-nano high aspect ratio structural weak interference images

Wide-spectrum interferometry is a highly applicable and precise measurement method, which can be used to measure the small-diameter micro-nano high aspect ratio structure of semiconductor devices. However, with the reduction of the diameter size of micro-nano structures and the deepening of depth parameters, the contrast of interferometric measurement signals has significantly decreased, and noise interference has become increasingly severe. To address this issue, a two-stage wavelet decomposition enhancement algorithm was proposed in this paper. The algorithm employed a two-level processing framework that combined adaptive global enhancement with Bayesian local denoising. First, based on the db5 wavelet basis, a global threshold was dynamically determined using information entropy to preliminarily improve the fringe contrast. Subsequently, on the db8 wavelet basis, local refined denoising was performed by integrating neighborhood energy analysis with the Bayesian threshold. Experimental results indicate that the error rates of depth measurement for high-aspect-ratio structures are all below 0.2% when the proposed algorithm is applied. Compared with conventional discrete wavelet transform, Bayesian denoising, and contrast-limited adaptive histogram equalization methods, the error rate is reduced by at least 12.3%. The algorithm is shown to effectively resolve the challenge of high-precision extraction of weak interferometric image information from micro-nano high-aspect-ratio structures, providing a reliable technical solution for high-precision non-destructive measurement of small-linewidth micro-nano structures with high aspect ratios.