Laser imaging radar has high space resolution, and can output four-dimensional (4D) image (intensity image+three-dimensional (3D) range image), so it is fit to be the sensor of target recognition. Support vector machine (SVM) is a way of target recognition, which has high correct recognition rate, in the case of learning with small samples. Through optimizing the algorithm of SVM, it is embedded into digital signal processing (DSP) of laser imaging radar, achieving the function of target recognition, which has high practical meaning. The experiments with the real intensity images of laser imaging radar are finished. In the experiments, the testing set totally has 56 samples, and the costing time is 31.97μs, which can state that the algorithm can meet the real-time requirement, and the correct recognition rate is 98.2%. Another experiments with simulated range images of laser imaging radar validate the generalization of SVM. The experiments results show that SVM can meet the recognition demands of laser imaging radar in the real-time and recognition performance.