Three-dimensional Morphology Measurement of Highly Reflective Objects Based on Three Channel Parallel Processing
Currently,optical 3D measurement technology is widely applied in various fields including industrial inspection and reverse engineering,due to non-contactness,high speed,and high precision.However,traditional fringe projection methods face challenges for the measurement of highly reflective surfaces.In such cases,the projected images may experience overexposure in bright regions,which may result in camera distortion.Currently,commonly used methods such as the multiple exposure time method and the intensity adjustment of projection patterns have issues such as requiring a large number of images to be captured and having poor algorithm adaptability.Therefore,to ensure both the measurement efficiency and accuracy while adapting to varying surface reflectivity of highly reflective objects has become a challenging research topic.With respect to the three-dimensional surface measurement of highly reflective objects,this investigation presents an approach for fringe projection intensity prediction and image fusion based on three-channel parallel acquisition of color camera.In the first place,a uniform gray image is pre-projected onto the object's surface.Based on the histogram distribution of light intensity of the object being measured,regions with similar surface reflectivity are grouped into clusters.The specific values for either three significant clusters or two significant clusters plus one minor cluster are calculated.By considering the scattering effects of light of different wavelengths,the optimal projection intensity ratio for the three channels specific to the object is determined.The corresponding RGB fringes are then re-projected onto the surface of the object.Secondly,the corresponding RGB fringes are re-projected onto the object's surface,and a color 3CMOS camera with three independent sensors captures images in parallel.Under multi-channel illumination conditions,the camera exhibits distinct responses to fringe patterns projected the projector with varying intensity ratios,subsequently isolating the corresponding channel-specific fringe images.Following this,pixels in positions with both unsaturation and maximum modulation are selected from the images of the three channels to generate image masks for each channel.Finally,based on the masks and fringe images from the three channels,a high dynamic range fringe pattern is synthesized.A hardware measurement system is constructed using a 3CMOS camera and a color DLP projector.Camera calibration is performed to obtain the internal parameters,and system calibration is carried out to acquire the geometric correspondence parameters between phase and space.The captured high dynamic range fringe images are processed using the four-step phase-shifting method and the optimal three-fringe selection method to calculate the phase and obtain unwrapped phase information.Utilizing the parameters obtained from the system calibration,the three-dimensional reconstruction is then completed.Using the proposed method,measurements were taken of highly reflective metal planar and spherical parts.The experimental results indicate that the method can adaptively predict the optimal projection intensity ratios for different highly reflective objects across the three channels of the color camera.Profile data from four locations on the two reflective objects were selected for fitting,and the root mean square error,maximum residual and correlation coefficient of the original data and the fitted data were analyzed.The measurement error of this method is reduced to 61.2%of the measurement error of traditional methods.In contrast to the multiple exposure time method and the adjustment of projection pattern intensity method,this approach utilizes only one pre-projected image and 12 fringe images,which greatly reduces the number of image acquisition in the process of multiple exposures or adaptive pre-projection,improves the measurement efficiency of the system,and has high measurement accuracy.In conclusion,the presented method may find promising applications in scenarios involving highly reflective metal components,and shows potential for industrial inspection and similar applications.
Measurement3D morphologyStripe projectionHighly reflective objectsThree channelsHigh dynamic range
万方数据
维普
