An infrared dim target detecting algorithm called the adjacent relevant pixels detection is proposed. Firstly, an adaptive local threshold is used to detect bright pixels, and then eliminate the noise with the information of adjacent relevant pixels. The remaining suspicious pixels are enhanced by the relevant bright pixels, by which the noise is suppressed again and considerable SNR gains are obtained. ARPD method is compared with classic TopHat transformations with and without 8neigboorhood clustering. The target point in images processed by ARPD method receives high signaltoclutter ratio gain, and the detectability of the target is enhanced.