Image Distortion Correction System And Its Engineering Applications

In the Factory production workshop, automated production has been gradually replacing the manual work. The pin distance detection of the electronic components (resistors, capacitors, etc.) is an important link in the automated welding process. Electronic components in the production process will becomes pin crooked and pin shortage.CCD cameras used in the factory are generally wide-angle lens.when wide-angle lens captures images, images are prone to radial distortion, especially barrel distortion. if the image suffered from the barrel distortion are not corrected, it will make the electronic components original pin distance changed.so the electronic components will be mistaken to be damaged so that it will be automatically discarded. thus it is crucial that the distance of the pin on the electronic components is detected in the automatic welding.After taking wide-angle lens,in order to restore an image as much as possible, you need to use the image distortion correction algorithm.Distortion correction mainly contains two type of methods. One is the template method and the other one is Parameters known method-equivalent spherical law. Accuracy of template method is high, but the method need to produce a template. So the computational complexity is big and it possesses bad real-time. The parameters known method is fast, but the accuracy is not high. The problem is that When we use the two methods,the template method need create a template template and parameters known method-equivalent spherical law need know the lens focal length or angle.To solve these problem, we use a strong applicability of the correction algorithm which it don’t face the specific lens.In the article,we use polynomial modeling to simulate the barrel distortion.First in order to ensure that the requirements of high-precision, sub-pixel edge detection technology is used to search for the images.then use the method of least squares linear fit..Finally the LM algorithm is used to solve parameter optimization of the sum of squared errors. the experimental results show that the accuracy of the calibration results are Within a precision of one pixel and better to meet the need of automated assembly.