Frequency domain feature-based component identification and positioning method

Abstract

The invention discloses a frequency domain feature-based component identification and positioning method. The method comprises the steps of performing continuation by taking alignment of an angular point of a component boundary in a to-be-detected image as a standard to obtain a pure-color continued time domain image; performing discrete Fourier transform on the time domain image to obtain a frequency domain image corresponding to the time domain image; obtaining a binary image corresponding to the frequency domain image; calculating the angle of the longest straight line in the binarized image; and calculating the angle deviation of the component according to the angle of the longest straight line. The angle deviation of the to-be-detected component is rapidly acquired based on frequencydomain characteristic image processing, the problems of large detection error and low efficiency caused by the fact that a detection result based on a time domain image is easily influenced by ambientlight and light change are solved, and the product yield is improved.

Description

technical field [0001] The invention relates to the technical field of visual recognition of flexible automatic production and assembly equipment, in particular to a component recognition and positioning method based on frequency domain features. Background technique [0002] As far as current personal computer usage habits are concerned, the keyboard is one of the indispensable input devices for inputting characters, symbols or numbers. Not only that, but all consumer electronic products in daily life or large-scale processing equipment used in the industry need to be equipped with a key structure as an input device to operate the above-mentioned electronic products and processing equipment. [0003] When intelligently assembling the keycaps of the keyboard, its positioning error must be less than 0.05 mm. The angle of the keycaps picked up by the mechanical arm from the feeding tray will have a relatively small deviation every time. This requires rapid identification and ...

AI Technical Summary

Problems solved by technology

[0004] In view of this, the embodiment of the present invention provides a component identification and positioning method based on frequency domain features to solve the problem that the detection results are easily affected by changes in ambient lighting and light in the prior art that uses time domain images for detection. , so that the detection error is large and the detection efficiency is low, which leads to low assembly efficiency and high product defect rate.

Images