Accuracy requirements, image analysis

More recently, image analysis has been employed to identify fibers and measure structural characteristics in nonwovens. However, the accuracy and indeed the limitations of these techniques have not been verified. A reliable evaluation of the algorithm requires samples with known characteristics. To accommodate this, a simulation scheme that generates specific nonwoven structures (images) has been developed. 

Jeong 2008) introduced the image analysis technique for finding the directions of warp and weft. He apphed the proposed method to some adorned fabrics like stripe, check, and dogtooth as well as imadomed fabrics of basic weaves to verify its efficiency and accuracy. (Kuo Su 2003) apphed gray relational analysis of the co-occurrence matrices to characterize the spatial properties of patterned fabric. However this approach is too sensitive to small variations of patterned texture and also requires intensive computation. 

In practice there are many ways in which we can attempt to achieve this and the technique selection will depend on the imaging modalities involved, the reason for coregistering the images and the accuracy of coregistration required for the analysis technique to be employed. 

---