An object image edge detection based quality-guided phase unwrapping approach for fast three-dimensional measurement

Abstract

Abstract²A new approach for achieving fast and reliable unwrapping of complex phase maps is presented for phase shifted based three-dimensional (3D) Fringe pattern profilometry (FPP). As phase unwrapping errors are often caused by shadows on fringe patterns images and object surface discontinuities, we propose to identify both of them. An image of the object is acquired by averaging the fringe patterns reflected from the object surface, based on which shadow areas and the phase discontinuities are identified. When carrying out phase unwrapping, the shadow areas and the edge pixels will be isolated, and then the remaining areas are unwrapped by means of an efficient technique proposed below. All the remaining pixels are then unwrapped by an efficient technique proposed, by which, the computational efficiency can be significantly improved. As proven by the experiment results, the proposed approach is able to implement fast and reliable phase unwrapping. I. INTRODUCTIONFringe Pattern Profilometry (FPP) has been an active area of research due to its potential application in many noncontact 3D shape measurement scenarios [1-3]. A typical FPP system consists of a projector, a digital camera and a computer. In most applications, the digital projector generates a group of sinusoidal fringe patterns images which are projected onto the object surface to be measured [1]. The image patterns are distorted due to the height variance of surface shape which are captured by the camera. As the fringe patterns are sinusoidal, and the deformed patterns can be modeled as the result of phase modulation of the projected patterns. Consequently, the fringe patterns can be described by phase maps, and the 3D shape can be retrieved by analyzing the phase maps. However, the obtained phases are wrapped into the range ( , ] SS referred to wrapped phases. In order to obtain the 3D shapes, the true phases are required and must be recovered from the wrapped ones. Such a process is known as phase unwrapping which is a critical step in the FPP approaches.Many phase unwrapping algorithms have been proposed. Existing phase unwrapping algorithms can be classified into two categories: the temporal algorithms (such as gray-code method [2] and multi-frequency method [3]) and spatial algorithms including simple path-following method [6], branch-cut method [7], and quality-guided flood-fill (QGFF) method [14]. The temporal algorithms are able to accurately recover the true phases from complex wrapped phase maps. However they usually require extra fringe patterns and thus not suit for fast 3D acquisition. On the other hand, the spatial algorithms can be applied in the cases of less number of fringe patterns, but they have limited ability to unwrap phase maps with isolated regions and discontinuities.

Phase unwrapping method based on parallel local minimum reliability dual expanding for large-scale data

“…Recently, improved schemes have been proposed, such as interwoven linked list, 5 priority queuing, 6,7 and quality mapping with edges and shadow areas. 8 Another implemented form of them, tiling strategies 9,10 based on residue clustering characteristics, have recently been provided with a convex hull algorithm for more effective schemes. The quality indices work more indirectly in reconstructing discontinuity compared to others based on residues, such as branchcuts methods.…”