Morphologic recovery method of product label noises based on image features and application in defect detection

To ensure the reliability of carbon fiber material products and eliminate all kinds of possible defects, it is necessary to take effective means to check their quality. X-ray based nondestructive testing technology combined with image recognition algorithm is considered to be a fast and efficient solution. However, the surface of the processed material is often accompanied by labels which contains various information. These labels are confused in the image and make most of the defect detection methods easy to misidentify. This paper mainly studies the recovery method of product label noises based on image features and application in defect detection. What's more, the proposed method can effectively and efficiently eliminate noises without affecting the rest of the image information which ensures the correct identification of defects in materials.