Dynamic timber cell recognition using two-dimensional image measurement machine.

Area of Science:

• Wood science and material identification.
• Image processing and computer vision.

Background:

• Image motion and defocus blur are common in capturing images of moving objects like timber.
• Low-quality images pose challenges for accurate timber cell recognition and species identification.

Purpose of the Study:

• To develop a robust timber cell recognition scheme for low-quality, blurred color images.
• To improve the accuracy of identifying timber species through cell recognition.

Main Methods:

• A novel two-dimensional image measurement machine with a color camera was developed.
• Image-moment-based blur invariant features were calculated.
• Timber cell recognition was performed using Euclidean distance based on moment invariants.

Main Results:

• The proposed scheme effectively utilizes image blur information to improve recognition accuracy.
• Experimental results validated the effectiveness of the blur invariant features for timber cell recognition.
• The study also theoretically and experimentally analyzed the maximum translation speed for image acquisition.

Conclusions:

• The developed scheme provides a robust method for timber cell recognition from blurred images.
• Accurate timber species identification is achievable, enabling correct assessment of timber properties and value.
• This technique enhances the ability to classify timber species based on cellular characteristics.