Motion perception is one of the most important aspects of the biological visual system, from which we get a lot of information of the natural world. In this paper, trying to simulate the neurons in MT (motion area in visual cortex) who respond selectively both in direction and speed, we propose a novel multiplicative inhibitory velocity detector (MIVD) model, whose spatiotemporal joint parameter K determines its optimal velocity. Based on the response amplitude disparity (RAD) property of MIVD, we build two multi-velocity fusion neural networks (a simple one and an active one) to detect the velocity of 1-D motion. The experiments show that the active MIVD neural network with a feedback fusion method has a relative better result.