計畫名稱 : 自動化鋼橋樑彩色塗裝鏽蝕影像評估方法
摘要:
近年來，由於電腦技術的發展讓影像處理技術具有實用性，不需花費昂貴成本即可將複雜的影像加以分析。早先針對鋼構橋樑表面塗裝鏽蝕程度評估所發展的影像辨 識方法，多將原來彩色影像轉換為灰階再進行處理程序。由計劃主持人和其研究團隊，探討一種自動化判斷模式，針對鋼橋樑塗裝鏽蝕狀 況直接以彩色影像進行辨認與量測，並提出一個基於 K-Mean 的影像分割技術，在 4 種 不同的色彩空間（RGB、HSV、L*a*b*、YIQ）下進行塗裝鏽蝕影像分割與量測，發現若將鋼構橋樑表面塗裝鏽蝕影像轉換成 Sb*組合進行 K-Mean 評估，與前人所提鏽蝕影像 辨識系統，如 NFRA、ICM、MPC，在處理效果上確實有長足的進步。然而這些方法在擷取數位化影像時，卻有部分限制，包括有：低對比數位影像及 畫面表層干擾等影像的辨識度仍無法有效提高。因此，此計畫最主要的目標，是除將色域理論應用於數位化彩色影像的處理過程， 並以多變量分析的分群理論來發展一個控制表層塗漆品質的自動化系統。鏽蝕程度評估系統將透過以下四個步驟來完成：
摘要:
近年來，由於電腦技術的發展讓影像處理技術具有實用性，不需花費昂貴成本即可將複雜的影像加以分析。早先針對鋼構橋樑表面塗裝鏽蝕程度評估所發展的影像辨 識方法，多將原來彩色影像轉換為灰階再進行處理程序。由計劃主持人和其研究團隊，探討一種自動化判斷模式，針對鋼橋樑塗裝鏽蝕狀 況直接以彩色影像進行辨認與量測，並提出一個基於 K-Mean 的影像分割技術，在 4 種 不同的色彩空間（RGB、HSV、L*a*b*、YIQ）下進行塗裝鏽蝕影像分割與量測，發現若將鋼構橋樑表面塗裝鏽蝕影像轉換成 Sb*組合進行 K-Mean 評估，與前人所提鏽蝕影像 辨識系統，如 NFRA、ICM、MPC，在處理效果上確實有長足的進步。然而這些方法在擷取數位化影像時，卻有部分限制，包括有：低對比數位影像及 畫面表層干擾等影像的辨識度仍無法有效提高。因此，此計畫最主要的目標，是除將色域理論應用於數位化彩色影像的處理過程， 並以多變量分析的分群理論來發展一個控制表層塗漆品質的自動化系統。鏽蝕程度評估系統將透過以下四個步驟來完成：
- 鋼樑橋表層塗漆影像擷取
- 色域最佳化
- 應用集群理論進行鏽蝕程度分群
- 鏽蝕程度評估 這個欲發展的系統，其驗證會確保質與量的並重。量的部份採用數理方法來推算鏽蝕的百分比；而質的部份則透過視覺化判定，及數位處理後之結果，判定是否能可 靠地反應彩色表層塗漆的影像。
Project : Automatic Color Image Assessment for Steel Bridge Rust Images
Abstrac:
Since computerized technology is rapidly developed, it makes the application of image processing feasible and low cost. Previously developed image segmentation methods for rust defect assessment do not directly process a color image. A color image was usually converted into a grayscale image for further processing.
The Principal Investigator (PI) and his team research into a computerized system for bridge surface quality assessment by means of color image processing. The new system can automatically identify and measure the steel bridge coating condition and defects through computers. This compares the performance of color image assessment for the rust images of steel bridge in four color spaces (RGB, HSV, L*a*b* and YIQ) by evaluating the seven color configurations for every single color space. The K-means algorithm is used to segment the rust area from the background. It shows that the rust images of steel bridge are directly converted into Sb* which is the combination of the component of the color spaces. Then the K-means algorithm is used to do segmentation. The new system shows better performance than previously developed image segmentation methods, like NFRA, ICM, MPC.
The system are often encountered difficulties while acquiring digital images under environmental conditions such as low-contrast digital images, and noises on painting surfaces. The purpose of this proposal is to explore the use of color space theories and clustering algorithm for processing digital color image and developing an automated system on controlling coating quality. The rust defect assessment system will be realized by passing through four steps: 1) steel bridge coating image, 2) optimal color space, 3) clustering of defects.4) assessment of defects.
Abstrac:
Since computerized technology is rapidly developed, it makes the application of image processing feasible and low cost. Previously developed image segmentation methods for rust defect assessment do not directly process a color image. A color image was usually converted into a grayscale image for further processing.
The Principal Investigator (PI) and his team research into a computerized system for bridge surface quality assessment by means of color image processing. The new system can automatically identify and measure the steel bridge coating condition and defects through computers. This compares the performance of color image assessment for the rust images of steel bridge in four color spaces (RGB, HSV, L*a*b* and YIQ) by evaluating the seven color configurations for every single color space. The K-means algorithm is used to segment the rust area from the background. It shows that the rust images of steel bridge are directly converted into Sb* which is the combination of the component of the color spaces. Then the K-means algorithm is used to do segmentation. The new system shows better performance than previously developed image segmentation methods, like NFRA, ICM, MPC.
The system are often encountered difficulties while acquiring digital images under environmental conditions such as low-contrast digital images, and noises on painting surfaces. The purpose of this proposal is to explore the use of color space theories and clustering algorithm for processing digital color image and developing an automated system on controlling coating quality. The rust defect assessment system will be realized by passing through four steps: 1) steel bridge coating image, 2) optimal color space, 3) clustering of defects.4) assessment of defects.