查詢結果分析
相關文獻
- 動態X光影像強化與瑕疵分類之研究
- 乳房X光影像中微鈣化之強化、特徵萃取及辨識
- Efficient Block Matching Algorithm for Video Coding Using the Generalized Edge Feature Matching Technique
- A Synthetic Wavelet Approach to Machine Vision Enhancement and Feature Extraction
- 化黑夜為白晝?--夜視鏡的功能與限制
- 數位電視ATSC-PSIP協定之應用設計
- 美國ATSC數位電視視訊標準之研究
- 影音壓縮大聖--MPEG-4
- 化黑夜為白晝--夜視鏡的功能與限制
- The Consistency and Reliability of Periodontal Bone Level Measurements Using Digital Scanning Radiographic Image Analysis--A Pilot Study
頁籤選單縮合
| 題 名 | 動態X光影像強化與瑕疵分類之研究=Study of Dynamic X-Ray Image Enhancement and Defects Classification |
|---|---|
| 作 者 | 江行全; 廖克東; 李淑惠; 王建智; 張子仁; 劉永賢; | 書刊名 | 工業工程學刊 |
| 卷 期 | 21:4 2004.07[民93.07] |
| 頁 次 | 頁409-421 |
| 分類號 | 494.568 |
| 關鍵詞 | X光影像; 動態影像; 低對比; 影像強化; 瑕疵分類; X-ray image; Dynamic image; Low contrast; Image enhancement; Defect classification; |
| 語 文 | 中文(Chinese) |
| 中文摘要 | 本研究主要目的為針對動態X-ray影像,發展一套即時影像強化與瑕疵分類系統。由於所欲檢測的產品組合材料複雜且需要透過旋轉進行檢測,導致經X光所擷取的影像不清晰,造成檢測誤判。對此,本研究首先將動態影像以一連串靜態影像的方式表示。然後針對所要檢測的產品A、B,利用最適強化指標分別選取出最適的強化演算法。最後,則提出以相關係數為基礎的物件搜尋演算法,配合馬氏區別函數分類法,進行產品瑕疵偵測與分類。將提出的方法透過C++ Builder 6.0撰寫執行程式,並與X光檢測機作連結,進行即時影像強化分析。經實驗分析,證實所建立的系統可以完全分別產品A、B。就影像強化而言,其對比值可分別提昇17%及40%。就瑕疵分類而言,其正確率分別為96%和100%。在實際線上測試,其所開發系統之平均檢測時間約為0.8秒。 |
| 英文摘要 | The objective of this paper is to develop a real-time image enhancement and defect classification of the dynamic X-ray image system. The inspected products consist of variety of materials and need to rotate the product for inspection, it raised more challenges for generating a good quality image in using X-ray. First, acquire a series of static image substitute for dynamic image. Then, we used optimal image enhancement index to select adaptive enhancement algorithm for products A and B, respectively. Finally, based on the correlation coefficient, the proposed object search algorithm and the combined discriminate function classifier to process defect detection and classification. In this study, C++ Builder 6.0 were used to develop computer software, and interfaced with the X-ray machine originally used in the workplace, to process real-time image enhancement analyses. The experiment analysis results showed that the proposed analysis procedure could identify different products completely. And, contrast value of image enhancement can increase 17% and 40%, respectively. The accuracy of defects classification have 96% and 100%, respectively. The average on-line inspection time is about 0.8 second. |
本系統中英文摘要資訊取自各篇刊載內容。