Research On Precision Measurement Technology Of Parts Based On Sub-pixel

With the development of social science and technology,the manufacturing industry's requirements for detection technology are gradually increasing.Traditional manual part size measurement is affected by subjective judgment and measurement tools,not only has low detection efficiency,but also has poor anti-interference ability.Non-contact inspection of parts through visual measurement technology is gradually gaining popularity.It is in the size measurement link.There are many advantages that cannot be compared with manual measurement,such as high precision,high efficiency,automation,etc.The sub-pixel-based vision measurement technology can obtain better measurement results than traditional pixellevel detection without increasing the system hardware resolution,reducing industrial costs.This article has carried out research on sub-pixel-based precision measurement technology of parts.The main research work is as follows:A nonlinear imaging model is established,and the mapping relationship between image edge points and object points is analyzed.Taking into account the distortion of the camera in each field of view,the tilt of the object surface and other factors,the internal and external parameters of the camera and the parameters of the object plane are calibrated,the image is geometrically corrected,and anisotropic diffusion filter is introduced to achieve the denoising and edge preservation of the image.It provides a guarantee of image quality for subsequent part size measurement.A sub-pixel image edge detection algorithm based on the combination of wavelet transform and Franklin moments is proposed.The algorithm uses wavelet function to decompose the target image at multiple scales to obtain image pre-edges,which are convolved with the sub-image of the Franklin moment template.Obtain the parameters of the sub-pixel edge step model,and calculate the accurate sub-pixel edge point position.And use this algorithm to carry out binary image sub-pixel location analysis,noise immunity experiments of different levels of noise,and edge extraction of images with rich texture details,and design evaluation experiments,and three commonly used sub-pixel detection algorithms.A comparative analysis was carried out to prove the feasibility of the algorithm.A part measurement platform was built.Firstly,a preliminary measurement experiment was carried out on standard measurement test blocks of different sizes,and the measurement results were analyzed and optimized.On this basis,the dimensions of industrial parts(hole caps,metal pads)were measured,considering In view of the complex environmental factors that the actual measurement system in the industry is affected by,a comparative experiment of different positions of the field of view is designed and the stability of part size measurement under the illumination range of 560lx?25000lx is analyzed.The light adaptability was evaluated.The experimental results show that: when the pixel equivalent is0.051mm/pixel,the detection algorithm proposed in this paper can measure the size of parts with a precision range of 0.43-0.71 pixel.In the case of the same computing speed,compared with the current mainstream algorithm with the higher accuracy,it has increased by about29.11%,which meets the requirements of actual industrial parts measurement.