In view of the truncated window causing the performance loss by the conventional sliding window decoding (SWD) algorithm to spatially coupled LDPC (SC-LDPC) codes, a SWD algorithm is proposed based on window extension (ESWD) in this paper to improve the reliabilities of the information in the decoding window, upgrading further the whole decoding performances. On this basis, a low latency with early termination window sliding scheme is proposed to reduce the number of sliding, realizing the good trade-off between the decoding performance and decoding latency. The simulation results show that the proposed ESWD algorithm has about 1.8 dB gain with decoding window size being small, compared with the conventional SWD algorithm. And with the increment of window size, though the performance gain decreases gradually, the performance is still superior to that of the sliding window decoding algorism at close decoding performance of the sub-optimal belief propagation algorithm. Moreover, the decoding complexity and latency expressions are also derived. The analysis results show that compared with the conventional SWD algorithm, although the slight increase of the decoding complexity in one decoding window is resulted by the additional few check nodes due to the window extension in the proposed ESWD algorithm, the whole decoding complexity keeps almost the same with the conventional SWD algorithm and the decoding latency is significantly reduced.