Surface topography data fusion of additive manufacturing based on confocal and focus variation microscopy

Additive manufacturing (AM) technology has become an important manufacturing tool owing to its outstanding capability in manufacturing parts with complex geometries and shapes with different materials. Importantly, surface texture is an essential feature of manufactured parts as it represents the evidence of a particular manufacturing process. In AM, the finished surface generally exhibits complex surface textures consisting of different marks like deep grooves and protrusions. Similarly, surface topography influences functional performance, like fatigue and wear behavior. Thus, understanding surface topography is essential in AM as it provides knowledge on the formation of surface marks during manufacturing and their corresponding effects on surface properties and functions.

Several techniques have been developed to facilitate precise AM surface metrology. Contact techniques use stylus tip to scan the surface area, followed by the reconstruction of the surface topography using the information of the tip displacement. In contrast, non-contact techniques are much faster, non-destructive and offer high resolution. Among them, coherence scanning interferometer (CSI), confocal laser scanning microscopy (CLSM) and focus variation microscopy (FV) are suitable for obtaining surface textures of surfaces produced via AM. CSI, CLSM and FV are mostly suitable for measuring smooth, rough and very rough surfaces, respectively. However, due to the highly complex nature of the AM surface textures, each of these three techniques experiences limitations in providing the desired results.

Recently, the focus has shifted to combining the advantages of individual techniques to obtain good-quality data sets. Additionally, using different objective lenses to conduct surface morphology and present multiscale surface textures has been considered a viable option. These concerns belong to the data fusion problem, with the former being competitive data fusion and the latter being corporative data fusion. Unfortunately, there are limited studies on data fusion in the field of AM.

On this account, Dr. Yibo Zou, Dr. Jiaqiang Li and Dr. Yusheng Ju from Soochow University proposed two innovative data fusion methods for reconstructing surfaces produced by directed energy deposition (DED) AM. The surface topographic data were measured using CLSM and FV techniques. The first method was based on competitive data fusion concept, and it aimed at improving the data quality by combining the advantages of the FV and CLSM techniques. The second method adopted the cooperative data integration concept and generated a single representation containing local details and global information. Their work is currently published in the journal, Optics Express.

The authors showed that both the presented fusion methods achieved satisfactory results. For the competitive fusion, the fused data preserved the FV data characteristics in the long wavelength areas. Its vertical resolution was also improved by integrating short wavelength areas from the CLSM data. These improvements resulted in eliminating the derived artifacts like spikes and ghost points. On the other hand, the cooperative data fusion achieved one-pixel surface registration precision by adopting the feature-based registration method aided by the color image formation. Compared with the conventional ICP algorithm, it was computationally faster and required and consumed less energy.

In summary, Soochow University scientists reported new data fusion strategies capable of fusing topographic height data obtained from CLSM and FV techniques used in DED manufacturing. While there are no universal laws determining the selection of concrete parameters for data fusion models, a general rule adopted in this study recommended a greater threshold value and a greater weight when fusing data obtained with an objective having a higher numerical aperture. In a statement to Advances in Engineering, Dr. Yibo Zou first author said that the presented data fusion strategies provided innovative solutions that would contribute to advanced surface representation and microscopic reconstruction in multiscale in the field of AM.