A method for constructing analog image library for visual alignment algorithm screening

Abstract

The invention provides a method for constructing an analog image library used for visual alignment algorithm screening, which comprises the following steps: selecting feature shapes of alignment; selecting a background image; setting a position of the feature shape in the background image, that is, a position of the image one constituting the feature shape in the simulated image pair; setting a displacement and a rotation angle of a feature shape in an image 2 in the simulated image pair with respect to a position in the image 1; the feature shapes are processed by the alignment feature processing algorithm to generate image one and image two of the simulated image pairs. The invention has the advantages that the analog image library constructed by the method comprises a large number of image pairs which may be placed at arbitrary positions and may be distorted by the feature shapes; feature recognition algorithm can be improved by using analog image database to find the shortcomings or defects of the algorithm, which can improve the matching accuracy of the algorithm and improve the robustness of the algorithm, and shorten the debugging time of the algorithm in the field application.

Description

technical field [0001] The invention relates to the technical field of visual alignment, in particular to a method for constructing a simulated image library for visual alignment algorithm screening. Background technique [0002] Alignment is a professional name for the precision assembly of devices in modern industrial production. Its typical applications are the installation of various flexible or rigid devices typically represented by mobile phone production. The specific implementation process is to install the object A at position 1 and the object B at position 2 together. During the installation process, the horizontal or rotational direction of object A or object B needs to be adjusted. A key link in realizing the alignment function is whether the precise positions of the above-mentioned object A and object B can be obtained. In order to achieve precise adjustment of the position, the object will be photographed through the visual alignment system during work, and th...

AI Technical Summary

Problems solved by technology

The advantage of the actual machine verification is that it conforms to the actual situation, but the disadvantages are also very obvious. It needs to be carried out on the actual production site, the joint debugging time is long, the cost of supporting hardware investment is high, and due to operational restrictions, abnormal situations such as failures of the verification machine The lower alignment software effect is often more difficult

Images