Automatic Classification and Recognition of Tourist Attraction Images using k-Means Algorithm and Image Processing Technology

This study investigates the automatic classification and recognition of tourist attraction images through an enhanced K-means algorithm paired with image processing techniques. Given the growing interest in efficient tourist image management and promotion, our research presents a system that significantly advances the processing and categorization of scenic spot images. We implemented a variety of image preprocessing steps—such as grayscale conversion, normalization, Gaussian filtering, and edge detection — to improve the accuracy of feature extraction. Our method involved an optimized K-means algorithm, modified to select initial centroids more effectively and adjust cluster numbers dynamically. Experimental results, based on a dataset of 150 images, demonstrated a classification accuracy of 92%, with the system maintaining robust performance even as dataset sizes increased, confirming the algorithm's efficiency and stability. The processing time for 5000 images was recorded at 12.3 seconds, showcasing a linear relationship between data size and processing time. These findings highlight the potential of our proposed system in enhancing tourism management and marketing strategies by providing a reliable and scalable image classification solution.