April 2007 / The Expoprotection Info / News smart cameras / After through outputs in Mpeg4 or H264,  PoE, very wide dynamic range sensors or  megapixel sensors (multi), cameras including an intelligent analysis of the source  image arrive on the market, answering the needs of faster occurence detection  and closer images.''Vincent Carrier”, Technical Director of Nexvision, has told us that he has chosen a partnership with Kaolab for the integration of image analysis algorithms, especially giving a motion detection really workable without false positive. Working on 5 Mpixel of  NexDome requires a lot of power, and Nexvision has provided a tool for working with images. Metadata has been used intensively, Kaolab having a large work experience  on statistics.Current developments go well beyond the simple motion detection, and use exchanges of parameters between the camera and the storage server. On  PC,  compressed images had to be brought, and then decompressed before working on them, involving a double conversion. Today, with the DSP integrated in the cameras, you can work directly and earlier on given raw data , which gives a much higher efficiency. We use in each camera a Da Vinci DSP of Texas Instruments, in its whole version, which includes a  C64 center of 600 MHz, an ARM processor and a SIMD, for a very high computing power. An important condition must be a very fast power transmission of data within the camera, and for this, there are huge differences from one processor to another. Texas Instruments itself makes very good data changeover matrix  in its DSP, and the choice of the Da Vinci had to be done for NexDome.