Electrode dirt detection method

Abstract

The invention discloses an electrode dirt detection method. The method comprises the following steps: preprocessing an electrode image to obtain an electrode grayscale image; according to templates inthe morphological operator template set, carrying out opening operation and closing operation on the electrode grayscale image to obtain an electrode foreground image and an electrode background image; performing autoquotient graph operation by utilizing the electrode foreground image and the electrode background image to obtain initial autoquotient graph data; 4, establishing an optimal auto-quotient graph data selection model, obtaining optimal auto-quotient graph data, and performing deviation standardization processing on the optimal auto-quotient graph data; performing thresholding processing on the standardized auto-quotient graph data to obtain a connected domain image; and extracting an outer contour of the connected domain image and filling to obtain mask data, and performing exclusive-OR operation on the mask data and the connected domain image to obtain a dirt segmentation result. By means of the invention, electrode dirt segmentation can be achieved, the adaptability to illumination changes is high, the workload of manual parameter adjustment is reduced, the calculation efficiency is high, and the detection precision is high.

Description

technical field [0001] The invention relates to the technical field of electrode detection, in particular to an electrode defect detection method. Background technique [0002] An electrode is a comb-shaped sensor element interspersed by the two ends of the electrode. Usually, the two poles of the electrode are passed into a square wave for reactance-based calculations. The electrodes need to be inspected and cleaned during production, but in some cases the electrodes cannot be packaged well during use, such as glass substrates, etc. Since ordinary dirt can absorb part of electromagnetic waves, once the electrodes come into contact with dirt without protective measures, problems such as short circuit and signal attenuation will occur. [0003] Currently, methods for detecting soiling include two types. One is designed around the MURA (display unevenness) detection requirements of liquid crystal panels and other products, and needs to provide environments with special requi...

AI Technical Summary

Problems solved by technology

The general detection equipment takes pictures directly, so the collected images are mostly different, resulting in the need to readjust the manual parameters, such as the notch position, etc. Moreover, if the image input size changes slightly, the notch position is likely to be different from the frequency spectrum. The response position of is biased, thus introducing other regular textures

In addition, DFT has certain restrictions on images with thick lines and strong edges such as electrodes. Generally, scenes with frequency domain filtering methods such as DFT have lower amplitude and higher frequency textures.

For similar components such as comb electrodes, due to the existence of obvious step boundaries and large brightness changes, it is difficult to find the frequency position corresponding to the texture in the DFT spectrum, and it is also difficult to completely eliminate all related frequency responses

Moreover, the electrode gradient changes more according to different models, resulting in different notch positions

Therefore, it is difficult to use the method of hand-designed notch filter to deal with

[0005] MURA and DFT methods based on dedicated light sources limit detection feasibility due to usage patterns

Therefore, the existing electrode element contamination detection technology has the problems of high requirements on the light source, heavy workload of manual parameter adjustment in the feature extraction method, large amount of calculation in the detection method, and low detection accuracy.

Images