Machine Vision for Sculpture Surface Optimization

The article under research is called Machine Vision of Sculpture Surface Optimization and it addresses the topic of applying modern computer vision and artificial intelligence methods to improve the accuracy and efficiency of sculptural surface refinements. Conventional manual procedures in surface optimization processes, are very subjective, time consuming and their procedures are usually restricted by inconsistencies in human perception. This paper suggests a powerful vision-based framework, which uses machine learning and 3D imaging technology to reach automated and data-driven surface analysis and fixation. The system architecture involves high resolution cameras and structured lightings modules and robotic manipulators to have accurate data acquisition and surface analysis. The system identifies the important geometric and texture features of the point clouds by preprocessing the image, edge and curvature detection, and analysis of the image. Convolutional neural networks (CNNs) and autoencoders are used to improve these features so that surface irregularities and deviations of the intended topography are predicted. An optimization loop that is based on a feedback loop is used to allow the correction of tool paths or polishing parameters in real-time. It has been proven in experiments that smoothness and uniformity of the surface, as well as reduction in defects, improved considerably under comparison to the manual methods of handling the problem, whereas processing time and human intervention were reduced significantly. The research also confirms the approach by use of visual inspection, 3D reconstruction metrics, and quantitative indices of quality of surfaces

Convolutional neural network · Image processing · Machine Vision · Point cloud · Polishing · Preprocessor · Advanced Computational Techniques in Science and Engineering · Advanced Theoretical and Applied Studies in Material Sciences and Geometry · Innovations in Concrete and Construction Materials