Research On Step Object Measurement Based On Binocular Data Fusion

Aiming at the shortcomings of traditional three-dimensional measurement methods in image processing such as complex structure,complicated process,lower precision and so on,a three-dimensional measurement method based on binocular data fusion is proposed.This method combines monocular and binocular cameras with line structured light measurements.That is fused to get more accurate data.And the system is built and tested for the object on the step surface.Compared with the traditional method,experimental results show that this method is more adaptive to the actual scene.The main research results as follows:1)System design based on blind area generated by object step surfaceAiming at the images that can be captured by the left and right camera,we should take “Average fusion” processing.For the left camera,the right camera can't get the image.The method of “complementary integration” is applied to this region.2)Improved light stripe center extraction algorithmAccuracy of central fringe extraction has great influence on image feature point extraction and matching.In the experiment,the gradient extremum method combined with barycenter method is used to extract the center of the light stripe.Selecting the appropriate threshold,and then extracting the gray center of gravity.3)Improved structured light image processingIn this paper,the author compares the median filter and mean filter to the structured light image processing.The author choose median filter to process two images.After that,a comparative test of various edge detection operators is carried out.A Canny operator with better processing effect on the edge of structured light stripe for complex surfaces is selected.The mature and widely used gray barycenter method is used to extract the light stripe center.The feature points are extracted and matched by complex matching based on SURF algorithm and Epipolar Constraint Principle.The threedimensional image of the object's stepped surface is obtained.