A new block-based recursive moving-target indication algorithm in velocity domain is proposed to solve the problem which is the rapid detection of dim and small target for infrared search and tracking system. Firstly, the two-dimensional least mean square filter is adopted to filter the infrared image sequence, which extracts small targets and residual errors of image sequence. Then, block-based recursive moving-target indication algorithm is adopted to accumulate small target in image block sequence for the enhancement of small target velocity in velocity domain. Finally, resulting image is obtained by using classical recursive moving-target indication algorithm and target velocity for small target detection. Compared with classical method, the proposed method requires less running time, and can be used to detect dim small target image effectively as demonstrated by several groups of experimental results.