Digital Revival of Traditional Batik Patterns: Accurate Extraction With Mask Attention AR-Net and AIGC Design Application

The growing need for digitally preserving the cultural heritage of ethnic minorities has positioned the accurate segmentation and innovative utilization of traditional patterns as a pivotal challenge in both technological and cultural preservation domains. To tackle this challenge, a novel method for segmenting low-resolution batik patterns (MAR-Net) is proposed, and its potential applications in cultural and creative product design are investigated. Specifically, an adaptive rectangular convolution (ARConv) mechanism is incorporated into the conventional U-Net architecture to effectively capture the intricate and diverse texture features of batik patterns, while the AR-Net network is developed to facilitate improved feature extraction and the integration of local and global information. To further enhance segmentation precision, a mask attention mechanism (MaskAttention) is utilized to strengthen the network’s focus on critical regions, thereby improving detail restoration and pattern segmentation through dynamic adjustment of feature weights. To improve the model’s robustness and discriminative capacity, a composite loss function is formulated, integrating pixel-level cross-entropy loss with instance-contrast-based attention loss to enhance the discriminative power of the feature space. In terms of application, accurately segmented batik patterns are converted into resources for cultural and creative product design, and a digital batik pattern resource library is created, with the potential for incorporating traditional patterns into the AIGC creation process being explored. Experimental results indicate that the proposed MAR-Net method not only markedly enhances precision in low-resolution pattern segmentation tasks but also exhibits substantial potential for application in cultural and creative product design, offering a novel solution for the digital preservation and innovative utilization of batik patterns, thereby significantly contributing to the study and preservation of traditional ethnic minority arts.