Application Research And Development On Vision Measuring System Of Shaft Parts Based On DM6437

Measurement is an important link to ensure quality of industrial products. With the standardization and legalization of product quality management, it is urgent to find out a method which is suitable for the mass, high precise and multi-parameter measurement tasks of all parts in the enterprise. However, the vision measuring system, a optical-electronic-machinery integrated system, is widely used in the industry, by right of the advantages such as non-contact, high speed measuring, high measure precision, continuous working and easy to realize automation, etc.This thesis combined national natural science foundation project and the application requirements of enterprise. According to the application bottleneck of the vision measuring system, such as measuring accuracy, measuring speed and economy, etc, this thesis makes a further research of designed and built embedded vision measuring system, selected and optimized image processing algorithm which include rapid and precise edge-detection, target line extraction and sub-pixel edge locating algorithm and designed automatic mechanism. The main works of the thesis include:1. On the basis of defining design task and comprehensive investigation, this thesis puts forward the design requirements and the design steps of vision measuring system and studies system design of embedded processing platform, optical imaging device, automatic mechanism and measurement process, etc.2. With DM6437 of TI Company based on DaVinci as the core processor, a image high-speed acquisition and real-time displaying system for shaft parts is completed after configuring the relevant registers in the video decoder-TVP5150 by I2C communication interface and resetting the registers in video processing subsystem. Meanwhile, embedded platform reduce the hardware cost comparing to the traditional PC and data acquisition card platform.3,The optimal image processing algorithms for DM6437 is proposed and realized by comparing and optimizing various algorithms based on MATLAB image simulation. According to evaluation criteria based on edge detection accuracy and speed, median filter, vertical sobel edge detection operator and 8 connected boundary tracking algorithm are proposed to get the single-pixel edge by the compared experiment of edge detection algorithm. In order to accelerate the extraction rate of target line and achieve the detection of scratches, a parameter optimized Hough transform algorithm based on geometrical property of shaft parts is proposed. In order to advance measure precision, an adaptive threshold corner detection based on curvature scale space is developed to get the start and end coordinates of the line, and least-squares fitted regression algorithm is developed to realize sub-pixel location of edge straight lines corresponding to parts of external diameter, height and taper.4,This thesis applies the above research results to the vision measuring system of perforating charge, and designs the feeding, conveying, positioning and sorting mechanism as well as the control processes. A vision measuring prototype of perforating charge is designed and built by considering part positioning errors and camera distortion error. The prototype is performed on perforating charge whose external diameter is 26.51mm, and the system's hardware resolution is about 0.04mm. The measurement results show that the average accuracy and repeatability of external diameter, height and taper can reach 0.016mm,0.034mm and 0.013, and the measurement speed is about 1.75 seconds/pieces. From the measurement results we can see that the prototype can reach at least 0.5 pixel accuracy. What's more, the prototype of the measurement accuracy, speed and economy are satisfying and achieving the task requirements.