Detection of Foreign Bodies and Bubble Defects in Tire Radiography Images Based on Total Variation and Edge Detection

We present the analysis and decomposition of tire radiography images by combining the total variation, curvelet transform based image enhancement, and Canny edge detection to detect foreign bodies and bubble defects in tires. Relying on the feature of total variation that images can be decomposed as texture parts and cartoon parts, we decompose the tire radiography image and select the cartoon part for defect detection since the textures are segmented and defect information is retained. The edges of the image are enhanced by modifying the curvelet coefficients before further edge detection operation. Furthermore, a Canny edge detection operator is used to detect the defects in which the eight-neighborhood bilinear interpolation non-maximum suppression method is employed to improve the detection performance. In our experiments, the Sobel operator and state-of-the-art methods such as the LoG operator and Canny edge detection algorithms are employed for comparison, and the experimental results are discussed briefly. The experimental results indicate that foreign bodies and bubbles in tires can be detected and located accurately by our proposed method.