A centralized hough track initialization algorithm is introduced for line in 3 dimensions space.Based on Roberts representation of 3D line

the hypothesis statistics is designed for Roberts parameters similarity.Through random sampling data from multiple sensors

the candidate Roberts parameter for 3D line is created corresponding to the data samples.Similar Roberts parameters are merged to accumulate to proper detection threshold for valid and credible line track.To reduce unreasonable random sampling and improper Roberts parameter merging

the object moving speed constrains are check when algorithm is realized

which will not only speeds up algorithm

but also decreases track false alarm ratio.Through performance analysis from simulation

our 3D centralized hough algorithm can be adapted to multi-sensor

multi-object environment

where different data updating rates

dense clutter and strong jamming will be encountered.