Single-stage 3D object detection based on pillar structures has gained attention for its high inference efficiency in autonomous driving. However, the quantization of point clouds into pillars often leads to the loss of fine-grained structural details, limiting performance in detecting small objects under sparse scenarios. To overcome this challenge, we propose an efficient LiDAR-based detection algorithm that incorporates multi-source feature encoding and structure-aware enhancement. Specifically, we design a Structure-Aware Feature Encoding (SAFE) module that integrates intra-pillar attention, inter-pillar structural modeling, and height histogram encoding to improve the geometric representation of pillar features. The backbone further incorporates a large kernel attention mechanism to capture long-range dependencies, along with an atrous spatial pyramid pooling module and a weighted dual-scale feature fusion module to strengthen semantic expressiveness and detection accuracy. The proposed method is evaluated on the KITTI dataset, with extensive visualizations and ablation studies. Results show an improvement of 3.9% in mean 3D average precision (mAP) over the baseline, including 3.66% and 4.59% gains for pedestrian and cyclist categories. The model runs at 34.2 FPS, demonstrating both accuracy and efficiency for real-time applications.