A network intelligent video target detection based on multimedia DSP

In this paper, a intelligent video camera based on TI DM6437 DSP which including hardware platform, software platform and algorithm realization is designed. In this system, the video data is received from the Sensor, the Target Detection algorithm is ran to detect objects. After that, the video data which include the information of the target detection is encoded and be transferred to the client pc by the cable. The JPEG, H.264 or MPEG4 format can be used to encode the data; In this paper the data is encoded by the H.264 format. On the client pc, the H.264 data can be decoded and the video including the detection information can be watched. The Mixture Gaussian Background Model and Markov Random Field are used to realize Target Detection and the Kalman Filter is used to filter the noise. By the tests, the Target Detection can be realized correctly in this system and the result can be watched on the client pc by using the GUI which is develop to decode the H.264 format data. Experimental result shows that this system is stable and fits real-time requirement.