Siamese Network Based CDC For Tile Color Difference Detection

Tile color difference refers to the inconsistency in surface color of tiles caused by factors such as raw materials or manufacturing processes, which negatively impacts product quality. Traditional tile color difference detection methods often rely on manual visual inspection, which is time-consuming and prone to subjective errors. To address this issue, the paper suggests a Siamese network structure utilizing central difference convolution. Two identical convolutional neural networks are employed to process reference and test samples, and key features are extracted using central difference convolution. To evaluate the effectiveness of the proposed approach, a large dataset containing numerous tile samples is constructed, and detailed experimental evaluations are conducted. The experimental results demonstrate significant performance improvements in tile color difference detection tasks using the Siamese network architecture based on central difference convolution. The outcomes of this study are of great significance in enhancing the level of quality management in tile production processes and provide a new approach to image-based color difference detection. Future research can further optimize the network architecture and algorithms, explore more feature representations and optimization strategies to further enhance the performance and practicality of tile color difference detection.