應用影像處理技術於鑽尖檢測自動化之研究

Research on Application of Image Processing Technology in Drill Tip Inspection Automation

李建興(CHIEN-SHIN LEE)

指導教授 ： 張元翔

中原大學/電機資訊學院/資訊工程學系/碩士(2023年)

https://doi.org/10.6840/cycu202301103

摘要

本文主要著重在使用影像處理方法，來降低不同待測品之間差異影響鑽針刃面檢測的穩定度與正確率，目前生產的微型鑽頭多半會使用鍍膜技術來降低鑽針進行鑽孔的損耗，鍍膜品質顏色各異，而再研磨針也常有殘膠、殘屑，或是暴露在空氣中過久導致的刃面氧化，這些不同的條件都會導致使用者需要針對各種樣品去調整檢測設定，使得自動化的檢測流程中，人為的因素增加，同時也會有檢測的判定標準不一致的問題。
成功的二值化應該要讓ROI(Region of Interest)保持適當大小，完整包含刃面的有效特徵，且過濾非預期的干擾。常見的刃面干擾因素有髒汙、亮度不足與反光，透過比較不同閥值之間的二值化處理效果可以觀察到，靜態的閥值設定，在標準亮度且無雜質的條件下，二值化處理成功率都在98.5%以上，然而當遇上鍍膜鑽針或是反光的狀況時，靜態閥值的二值化處理成功率卻不足60%，透過動態調整閥值的運算，建構可以自動過濾雜質，保留完整刃面影像的演算法，降低人為調整的差異性，簡化操作的難易度，也提高檢測標準的一致性，可以達到全樣品二值化處理成功率98.5%的表現。

關鍵字

影像處理 ； 微型鑽針 ； 動態閥值 ； AOI

並列摘要

This article mainly focuses on the use of image processing methods to reduce the impact of differences between different test items on the stability and accuracy of drill edge detection. At present, most of the micro drills produced will use coating technology to reduce the loss of drills for drilling. The coating quality and color are different, and the reshape drill often has residual glue, debris, or blade surface oxidation caused by being exposed to the air for a long time. These different conditions will cause the user to adjust the detection settings for various samples. So that in the automated detection process, human factors increase, and there will also be problems of inconsistent detection criteria.
Successful binarization should keep the region of interest at an appropriate size, fully contain the effective features of the blade surface, and filter unexpected interference. Common interference factors on the blade surface include dirt, lack of brightness, and reflections. By comparing the binarization effects between different thresholds, it can be observed that the static threshold setting, under the condition of standard brightness and no impurities, the binarization The success rate of binarization processing is above 98.5%. However, when encountering coated drills or reflective conditions, the success rate of binarization processing with static thresholds is less than 60%. Through the calculation of dynamically adjusted thresholds, the construction can automatically The algorithm of filtering impurities and retaining the complete blade surface image reduces the difference of human adjustment, simplifies the difficulty of operation, and improves the consistency of detection standards, which can achieve the performance of 98.5% success rate of binarization processing of all samples.

並列關鍵字

Image processing ； Micro drill ； Dynamic threshold ； AOI

參考文獻

1. 楊東樺, ”印刷電路板鑽頭之田口品質分析,” 碩士論文, 國立高雄應用科技大學, 2009.

2. R. C. Gonzalez and R. E. Woods, Digital Image Processing, Third Edition, Prentice-Hall, Inc. pp. 760-783, 2016.

3. N. Otsu, “A threshold selection method from gray level histogram,” IEEE Transaction on Systems, Man and Cybernetics, vol. 9, pp. 62-66, 1979.

4. M. Moganti and F. Erçal, "Automatic PCB Inspection Systems," IEEE Potentials, Institute of Electrical and Electronics Engineers (IEEE), pp. 6-10, 1995.

5. Lakshman Hazra, Hideo Kato, Takayoshi Kiryu, Yoshiyuki Hashimoto, “Inspection of reground drill point geometry using three silhouette images,” Journal of Materials Processing Technology, Vol.127, pp. 169-173, 2002.