Research On Three-dimensional Shape Measurement Technology Based On Color Phase-shifting Fringe Projection

With the continuous innovation of information technology and intelligent technology,collecting three-dimensional(3D)information of objects has gradually become an important way for human to understand and perceive the world.Unlike image information,3D information is used to quantify the shape and distance of an object.In recent years,3D measurement technologies based on different measurement principles have emerged gradually to meet the increasing demand for3 D information collection in production and life.The real 3D data of objects and the distance between objects can be obtained by 3D measuring equipment or 3D measuring technology.Currently,3D topography measurement technology has been widely used in computer vision,manufacturing inspection,mechanical engineering,biomedical engineering,cultural relic restoration,entertainment,and other fields.Due to the advantages of high precision,high resolution,non-contact and non-destructive measurement,fringe projection profilometry based on digital projection technology has become one of the important means of obtaining 3D shape data of objects.Based on the understanding and cognition of 3D measurement technology of structured light and digital projection technology,this dissertation focuses on the 3D shape measurement of dynamic objects and isolated objects based on color phase-shifting fringe projection technology,which can be summarized as follows:1.A single-shot color fringe projection 3D measurement method based on HSI(hue,saturation,and intensity)color space is proposed.This method adopts the color encoding method of the traditional color phase shift fringe projection profilometry(CPSFPP),by encoding the three-step phase-shifting grayscale fringe patterns into the RGB(red,green,and blue)color channels of a color image respectively to generate the color-encoded fringe pattern for projection in measurements.The captured color deformed fringe pattern is converted from RGB color space to HSI color space,and the corresponding point with the same hue in the reference hue is found by using the distorted object hue value,to calculate the fringe displacement of the distorted fringes.Then,the wrapped displacement is unwrapped to be the continuous fringe displacement,meanwhile,the intensity component in HSI space of the color deformed fringe pattern is used to calculate the binary template for selecting the valid region of the unwrapping.Finally,the height information of the object is constructed by height calibration.The proposed method avoids the process of the correction for the projector nonlinearity and compensation for color crosstalk in CPSFPP,which saves a lot of time and labour.The simulation and experiments verified the feasibility and effectiveness of the proposed method.The experimental results also showed that the proposed method can be used for 3D measurement of dynamic objects.2.The phase error due to the combination of system nonlinearity and color crosstalk is analyzed by a mathematical model in the CPSFPP.In addition to the gamma nonlinear effect of the projector and the color crosstalk effect,we also consider the nonlinear effect of the camera on the phase.An active phase error suppression method based on hue pre-correction is proposed to address this error.Firstly,the response relationship between the projected and captured hue is established,accodding to which the hue value of the projected color fringe pattern generated in the computer is modified actively,and the hue value of the captured distorted color fringe is corrected to be linear.The modified projected fringe pattern is projected to the object surface,and three phase-shifting fringe patterns decoded from the color deformed captured fringe pattern can be directly used for the 3D shape reconstruction of objects.Experimental results showed that the proposed method can effectively and actively suppress the phase errors caused by the combined effects of system nonlinearity and color crosstalk.The proposed method does not need any post-processing correction for the phase error and has a certain application potential in the measurement of dynamic objects.3.In order to reduce the number of fringe patterns required by temporal phase unwrapping(TPU)method in the 3D measurement of isolated objects,a high-accuracy 3D surface measurement method using hybrid multi-frequency composite-pattern TPU method is proposed.Firstly,a unit-frequency ramp pattern is combined with the low-frequency three-step phase-shifting fringe patterns to form three composite phase-shifting fringe patterns.The three composite fringe patterns are together with the high-frequency three-step phase-shifting fringe patterns to calculate three groups of wrapped phase at different frequencies.The high-frequency wrapped phase is unwrapped using the multi-frequency TPU method to obtain continuous phase,and then 3D shape of the object can be reconstructed.In addition,the effect of the system noise on the phase error in the proposed method is analyzed,and optimal frequency is estimated based on the analysis result to ensure the reliability of phase unwrapping and the high accuracy of measurement.Finally,the high-accuracy 3D measurement of the discontinuous or isolated objects is achieved by using grayscale hybrid method and color hybrid method.Experimental results showed that the proposed grayscale hybrid method can achieve the measurement accuracy as high as the traditional multi-frequency TPU method,and only requires 6frames of images.The accuracy of the proposed color hybrid method is slightly lower,but the advantage is that only 3 frames of color images are needed.4.In order to enlarge the measurement depth range based on the multi-view geometry-constraint based 3D shape measurement,a high-frequency color-encoded fringe projection profilometry is proposed.Six binary codes are used to encode two gray π-shift sinusoidal fringe patterns into the three RGB color channels of a color image,respectively,to genearate two color phase shifting fringe patterns.The wrapped phase and six color codes extracted from two color deformed fringe patterns,are used together with geometry constraints to match the correct corresponding points among the six candidate points in the measurement range for the 3D shape reconstruction of objects.The effect of the color crosstalk on the wrapped and the color code identification is analyzed in theory.The results show that the phase calculation is independent of the color crosstalk,and the color crosstalk has an effect on the hue value,however,the color code can still be identificated correctly by using proper threshold segmentation and color code boundary correction.With the use of color fringe patterns,the amplitude of the pattern is not reduced,and measurement accuracy is thus not degraded.The experimental results showed that the proposed method can effectively measure the 3D shape of isolated objects with only 2 frames of high-frequency color-encoded fringe patterns,and achieve stable and highaccuracy measurements in a large depth range.