In the orthogonal frequency division multiplexing(OFDM) system with IQ imbalances, a time domain least square (TD-LS) algorithm is proposed to estimate channel and IQ-imbalance parameters. To further exploit the sparsity of wireless channel, iterative shrinkage (IS) and parallel coordinate descent (PCD) are combined with the proposed TD-LS to form a new joint channel estimator TD-LS-IS-PCD. Simulation result shows that, in terms of bit error rate (BER)， the proposed TD-LS and TD-LS-IS-PCD algorithms perform much better than the traditional frequency domain least square (FD-LS) when the same least square equalizer is adopted. Furthermore, the BER performance of TD-LS-IS-PCD approaches that of the ideal channel estimator.