摘要： Roughness and texture analyses are an important topic for process development of high efficiency solar cells. Especially, strong improvement in Jsc was obtained by proper texturing of c-Si solar cells, which helped enhancing its infrared response. Usual pyramid size analysis relies on microscopy technique followed by image processing. One of the critical challenges to obtain an appropriate pyramid size distribution is to remove the initial contribution of background roughness, which is usually done by subtracting a polynomial background. This method holds strong limitation on physical meaning since the arbitrary chosen degree of the polynomial background will influence size distribution and can even provoke large errors. In this work, we present a method allowing to separate the contributions of different spatial frequencies and to filter the ones that need to be studied by applying Fast Fourier Transform (FFT). MountainsMap? software was used to study the roughness through a FFT treatment. A FFT spectrum was obtained in which selected frequencies were extracted to remove the noise and interferences from the initial confocal image. The resulted filtered topography image was obtained by the inverse FFT and analyzed to obtain the structural, morphological and geometrical characteristics of the surface. This methodology – applicable to several characterization methods – is particularly efficient when used with laser confocal scanning microscopy and opens the way to fast, automatic, accurate and non-destructive analysis that can be repeated at different positions of a sample with high throughput to enable statistical analysis.