Single-sensor imaging systems are widely deployed in portable devices including digital cameras, smartphones, and personal digital assistants (PDAs) for real-time image acquisition. While convolutional neural networks (CNNs) have demonstrated exceptional capabilities in various image processing tasks, their potential for demosaicking applications remains underexplored. This paper presents a demosaicking framework utilizing a Deep Residual Convolutional Neural Network (DRCNN) architecture. Firstly, we initialize the mosaicked images using conventional demosaicking algorithms and learn the DRCNN for three color channels. The proposed DRCNN architecture innovatively integrates three core components: Binary Convolution Units (BCUs) for computational efficiency, Efficient Layer Aggregation Networks (ELAN) for multi-scale feature fusion, and Dense Residual Blocks (DRBs) for enhanced gradient flow. Comprehensive evaluations demonstrate that the proposed algorithms outperform existing approaches in PSNR, computational complexity, and visual quality.