Flexible piezoresistive sensor array has broad application prospects in human-computer interaction. However, due to the complexity of reality, it is difficult to balance flexibility and perceptual ability in the process of tactile perception. Presented herein is a 4 × 4 matrix of a piezoresistive tactile sensor (TS) that is both pliable and composed of a porous blend of multi-walled carbon nanotubes (MWCNTs) and polydimethylsiloxane (PDMS). This sensor matrix is endowed with characteristics like pliability, consistency, and acute sensitivity, which facilitate its adherence to various shaped surface profiles. It boasts a peak sensitivity of 0.6 kPa−1 and is capable of detecting pressures within a broad spectrum from 0 to 640 kPa. An in-depth examination has been undertaken to assess the TS array's response to pressure, encompassing aspects such as hysteresis and repeatability. In addition to this, a scanning system for the array has been constructed to promptly detect, digitize, and present the pressure applied. A neural network model for three-dimensional force decoupling has been established to analyze the real-time data emanating from the sensor matrix, thereby enabling the precise forecasting of the three-dimensional force exerted upon the array.