Splicing Method Based On The Three-dimensional Surface Of The Marker Constraint

Digital morphology of large objects and closed surface can be realized by means of vision measurement, and it often needs to block measurement, the regional data is integrated by using three-dimensional image mosaic technique, realizing the overall morphology measurement of objects. Therefore, the three-dimensional image mosaic technique is a key technology to realize the object morphology measurement.According to whether there have constraints, image mosaic technique can be divided into free mosaic and constraint-based mosaic. Free mosaic extracts the characteristic information to realize the image mosaic through the common part of the adjacent detection area, suitable for checking the obvious characteristics objects due to its flexible. However, several times of mosaic prone to cause cumulative errors. Constraint-based mosaic can be divided into sensor location constraint and image control point constraint according to the different ways of constraint. The sensor views the object from a different perspective, the relationship between two adjacent sensors'location is known, this approach often requires ancillary equipment to calibrate the relationship of sensor location, and the operation is more complicated. Image control point constraint can realize the integration of image data by using the tag information, which was added to the measurement object, it just depends on the tag information to carry out the space image mosaic, but the extraction of tag information is very complex, and the mosaic errors are larger, so it's suitable for applications of less demanding to the mosaic accuracy.As a foundation of three-dimensional image mosaic technique, the accuracy and robustness of the three-dimensional detection restricts the precision and success rate of the three-dimensional mosaic. The three-dimensional detection method based on the structure light of symbols M array was put forward by the research group, the detection of dynamic scene and three-dimensional reconstruction was realized under the projection of binary light. This method is used in certain occasions, it can measure the object for a single color, and the surface curvature must small, the angle among the detection objects, the projection devices and CCD must small, and the light must weak. Due to the characteristics of detection objects, which have the small surface curvature, free mosaic is not suitable.Subject to the three-dimensional detection method based on the structure light of symbols M array that was put forward by the research group, a novel approach for three-dimensional image mosaic based on the mark constraint is put forward in this paper, which is based on the improvements of three-dimensional detection method. The method is set a group of markers on the detective object in order to get the marked image coordinate through extracting the markers, and then the marked world coordinate can be got, the transformation matrix can be solved by the world coordinates of markers, and then the three-dimensional image mosaic can be realized. This paper includes the following aspects:(1) Improve the three-dimensional detection method. The correctness of the calibration data is inspected, and the correction calibration data method is proposed. The windowing extension matching method is proposed according to the phenomenon of the decoding symbol mismatch, which effectively solves the problem of symbol mismatching. A smooth method which is suitable for the reconstruction surfaces of structured light three-dimensional detection is studied, and the problem of large burr can not be smooth in the reconstruction surfaces is solved, so has laid the foundation for the following mosaic work.(2) Study on marker extraction and matching method. Considering the marker plays a decisive role in three-dimensional mosaic, the marker extraction through the Golden Section methods to search threshold is studied in order to ensure there is no marker loss. Although this method's efficiency is not high, it's very stable. The world coordinates of the markers can be solved through the extracted markers image coordinates. Marker setting restricts marker matching and the transformation matrix solution, Marker setting is studied from the point of view of symbol matching, and the transformation matrix solution is also considered.(3) Study on mosaic method. The solving process of transformation matrix is introduced; the coordinate system of the mosaic image is integrated into the reference image coordinate system through the transformation matrix. The method of extract the common region of two images under the reference image coordinate is studied. Whether the two images registration is proposed according to the world coordinates of the detective points in the common region. According to the image without registration, the method of calculates the angle and displacement between two images is studied; the two images'registration was finally realized through the rotation and translation of the mosaic image many times.(4) The experiment of three-dimensional image mosaic. Image mosaic experiment was carried out, and the results of the experiment and the errors were analyzed.