The forthcoming multimedia telecommunication services are expected to use pre-encoded videos for storage and transmission. The heterogeneities of the present communication networks and the clients' devices demand matching the encoding format of the video source to the constraints of the networks and clients' devices. In order to provide quality support services to heterogeneous clients or transmission channels, the video server should have the capability of performing heterogeneous transcoding, which is regarded as a process of converting a previously compressed video bitstream into another bitstream with a different format. However, much investigation has been conducted to focus on homogeneous transcoding. Only a limited number of methods have been proposed to realize the heterogeneous transcoding from MPEG-2 to H.263. The major difficulties for transcoding B-picture to P-picture is that the incoming DCT coefficients arc predicted from the forward and backward prediction. In other words, the DCT coefficients predicted from the previous frame is not available. Therefore, the required new prediction errors need to be reestimated in the pixel domain. This process involves high computational complexity as well as introduces reencoding errors. Motivated by this, we propose a new transcoder architecture to convert a B-picture into a P-picture by making use of the techniques of motion compensation in the DCT domain and indirect addition of DCT coefficients. In this paper, we derive a set of equations and formulate the problem of how to obtain the DCT coefficients. Experimental results show that the proposed video transcoder achieve a better performance as compared to the conventional video transcoder in terms of both quality and complexity. Furthermore, we propose a fast algorithm to speed up the transcoding process making use of the correlation of motion activities between pictures.