In the field of intelligent tactile perception, achieving synchronous and precise monitoring of pressure and friction forces represents a fundamental challenge for replicating authentic tactile interactions. This complexity primarily stems from the intricate signal coupling between pressure and friction modalities. Drawing inspiration from the biomechanical mechanisms of human fingerprints, a dual-mode bionic fingerprint tactile sensor (BFTS) is developed that generates distinct capacitive responses to both pressure and friction stimuli. The sensor demonstrates remarkable pressure sensitivity, enabling precise discrimination of 3D blocks with varying hardness levels. Furthermore, its superior friction-sensing capability achieves accurate differentiation of 2D fabric surfaces with texture variations. To address the inherent signal coupling in concurrent pressure-friction detection, a hybrid deep learning architecture is devised, synergistically integrating Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and Attention Mechanism (AM). This multimodal fusion model achieves exceptional signal decoupling performance (R2 ≥ 0.95) through spatiotemporal feature extraction and adaptive weight allocation. Implemented on the BFTS platform, the integrated intelligent tactile perception system (ITPS) attains 97% classification accuracy for six visually similar citrus varieties. The proposed methodology not only resolves long-standing challenges in tactile signal decoupling but also establishes a new paradigm for multimodal perception in next-generation intelligent tactile systems.