Defect Detection in Cigarette Cartons Using Fast and Simple Projection and Difference Imaging Methods

To enhance defect detection in cigarette cartons, we employ fast and straightforward projection methods combined with difference imaging techniques. The projection method involves calculating horizontal and vertical projections of the image intensity values, which effectively maps objects onto different planes for improved contour and shape analysis. This can be implemented using sum projections along rows and columns with functions like numpy.sum() in Python or regionprops in MATLAB. The difference imaging method detects defects by computing pixel-wise differences between reference and test images, typically using absolute difference operations (cv2.absdiff() in OpenCV) or normalized correlation methods. These complementary approaches can be integrated through logical operations or weighted combination algorithms to comprehensively identify various defects such as size inconsistencies, color variations, and shape abnormalities. The implementation typically involves preprocessing steps like grayscale conversion, noise reduction filters, and thresholding operations before applying the core detection algorithms.