Compressed Sensing method suppress the MRI acquisition time for considering the patient's health. So the sensing process is carried out by a way of projecting an under-sampled data from spatial and k-space domain simultaneously. In this paper, our proposed Multi-domain reconstruction net acquires the under-sampled data with local binary pattern at the different sampling rates and reconstructs the resultant under-sampled data through the MD-USLBPRNET. Our proposed MD-USLBPRNET consists of two parallel channel and act together sections and also execute on spatial and k-space domain data simultaneously. Experimentally, the proposed method shows the performance better than the existing Deep learning method with the qualitative metric such as Peek-Signal-to-Noise Ratio and Structural Similarity Index.