Multimodal Aspect-Based Sentiment Classification (MASC) aims to identify the sentiment polarity of aspect terms in text by leveraging both textual and visual modalities. Although recent methods integrate detailed visual features, they often overlook the challenge that short and sparse multimodal tweets provide limited context for aspect semantics. Caption generation has been explored, yet such descriptions are usually generic and lack task-specific relevance. To address this, we propose a Knowledge-Enhanced Network (AKENM) that enriches missing context with external knowledge. AKENM comprises unimodal feature extraction, knowledge feature extraction, knowledge-guided interaction, and cross-modal fusion modules. By incorporating knowledge-enhanced textual and visual features, our approach better captures aspect-specific semantics and improves sentiment classification. Experiments on Twitter-2015 and Twitter-2017 show that AKENM outperforms competitive models in both accuracy and F1 score.