Invalid phase values removal method for absolute phase recovery.

A novel approach is presented for more effectively removing invalid phase values in absolute phase recovery. The approach is based on a detailed study involving the types and cases of invalid phase values. Meanwhile, some commonalities of the existing removal algorithms also are thoroughly analyzed. It is well known that rough absolute phase and fringe order maps can very easily be obtained by temporal phase unwrapping techniques. After carefully analyzing the components and fringe order distribution of the rough fringe order map, the proposed method chiefly adopts an entirely new strategy to refine a pure fringe order map. The strategy consists of three parts: (1) the square of an image gradient, (2) subregion areas of the binary image, and (3) image decomposition and composition. In combination with the pure fringe order map and a removal criterion, the invalid phase values can be identified and filtered out from the rough absolute phase map. This new strategy not only gets rid of the limitations of traditional removal methods but also has a two-fold function. The paper also offers different metrics from the experiment to evaluate the quality of the final absolute phase. In contrast with other removal methods, experimental results have verified the feasibility, effectiveness, and superiority of the proposed method.