Mask Image Generation for Tilted 3D Surfaces

The patent divides the single virtual image acquisition into multiple segments by generating multiple virtual images at different postures. The three-dimensional data is processed at multiple specific postures to create multiple virtual images, each capturing different portions of the surface. This segmentation allows the inspection system to cover the entire surface including previously hidden regions, resolving the contradiction between simplicity and accuracy.

The patent introduces a new dimension by adding posture variation to the virtual image generation process. Instead of generating a single virtual image at one fixed posture, the system generates multiple virtual images by changing the posture of three-dimensional data across multiple specific orientations. This dimensional expansion from single-point to multi-point coverage enables accurate mask image generation for tilted curved surfaces.

The patent applies preliminary action by pre-calculating and storing multiple virtual images at different postures before the actual inspection process. The three-dimensional data is processed in advance to generate a library of virtual images corresponding to various postures. During inspection, the system simply retrieves the appropriate pre-generated virtual images rather than computing them in real-time, significantly reducing processing time while maintaining complete inspection coverage.

The patent uses partial action by generating virtual images at a specific set of predetermined postures rather than continuously varying all possible angles. By selecting key representative postures from the full range of motion, the system achieves sufficient coverage of inspection regions without the excessive computational cost of generating images at every possible angle and posture combination.