Research And Implementation Of Sub-pixel Edge Detection Based On Partial Area Effect Model

With the development strategy of made in China 2025 put forward,the traditional processing and manufacturing industry is gradually transforming and upgrading to the direction of informatization,intelligence and automation.As an important technology for modern manufacturing industry to realize the integration of production and inspection and improve the manufacturing efficiency and product added value,visual measurement technology has been widely used in practical production.Vision measurement technology takes the collected image of the tested object as the detection target and information carrier,and obtains the required information through image processing algorithm.In the image,the edge contains rich visual information,which is one of the most important features of the image.Edge detection is to extract the location information of the edge from the image.The high-precision acquisition of the edge location is the prerequisite for many high-level applications,such as defect detection,target classification,size measurement and so on.In the practical application of vision measurement technology,the high-precision measurement results come from the high-precision acquisition of edge position.However,improving the performance of hardware equipment to improve the accuracy of the measurement system will lead to a significant increase in the hardware cost of the system.The edge position detected by traditional edge detection algorithm is based on the whole pixel in the image,which can not meet the accuracy requirements,while sub-pixel edge detection algorithm can calculate the exact position of the edge in the pixel,which can effectively solve the spear shield between detection accuracy and system cost.The main contents of this paper are as follows:In the beginning,this paper analyzes the noise distribution in the digital image,introduces the image filtering method,improves the integrity of the contour edge in the image through mathematical morphology operation,and effectively reduces the negative impact of noise and speckle information on edge detection.The principle and characteristics of edge detection operators in differential operator and optimization operator methods are studied.The most suitable pixel level edge detection operator is selected according to the specific application conditions through experimental comparison.The single pixel edge is obtained by fast thinning algorithm,which avoids redundant calculation in non-edge position of subsequent algorithms.Secondly,the principle and implementation of sub-pixel edge detection algorithm based on partial area effect model are studied in depth.Gaussian function is used to improve the calculation method of gray intensity on both sides of the edge,and a step-by-step sub-pixel edge detection method combining pixel level rough detection and sub-pixel fine detection is realized.The experimental results show that the accuracy of the method is 0.095 pixels on the synthetic image.Then,by comparing the operation time of each algorithm in the experiment of detect speed,the advantages of the step-by-step sub-pixel edge detection method are proved.Thirdly,the auto focus technology based on image processing is studied.Based on the traditional Brenner image definition evaluation function,a high pass filter template is introduced to improve the function's image definition evaluation performance and reduce the calculation.The experimental results show that the improved Brenner has good unbiasedness and faster computing speed,and the auto focus system based on it can ensure the good imaging quality of vision measurement system.Fourthly,the experimental platform of the vision measurement system is built,the focusing and distortion correction before the image acquisition are completed,the pixel equivalent is calculated by the secondary calibration method,and the calculation formula of the physical size is determined by the image of the measured object,which reduces the system error during the measurement,and increases the accuracy of the measurement result.At last,in the built of measure-experiment platform,taking the rectangular standard workpiece as the detection target,the size measurement experiment is carried out.The experimental results show that the step-by-step sub-pixel edge detection algorithm proposed in this paper has better detection effect,measurement accuracy and operation speed than the traditional sub-pixel edge detection algorithm.The measurement results are stable and can meet the high-precision requirements of the vision measurement system.