Research on Cigarette Defect Detection Method Based on G-KNN Algorithm

To prevent defective cigarettes from being mixed into packaging during the production process, this paper proposes a scheme for detecting defective cigarettes during production. Firstly, the Canny operator is utilized to perform precise edge detection on cigarette images, ensuring clear representation of the cigarette contours. Subsequently, a combination of Local Binary Patterns (LBP) and Gray Level Co-occurrence Matrix (GLCM) is employed for feature extraction, comprehensively capturing the subtle textures and structural variations on the cigarette surface. Finally, an improved G-k-Nearest Neighbors (KNN) algorithm incorporating a Gaussian kernel function is proposed for classifying the cigarettes. The results demonstrate that, compared with single feature extraction methods and the traditional KNN algorithm, this method significantly enhances detection accuracy to 0.96. The proposed method is more adept at handling complex texture features, nonlinear relationships, and subtle differences among samples in cigarette images, which is of great significance for improving production quality in the cigarette industry.