 Hello and welcome to this video about ST's new smart camera system designed to help customers develop secure and advanced automotive camera applications. Our smart camera based advanced driver assistance system is the perfect solution for a compact, low component count and low energy consuming camera system for automotive applications. Based on a high dynamic range CMOS image sensor and its companion image signal processor, this smart camera system is fully adapted to rear view, surround view, mirror replacement or any other analysis and tracking solution for around the vehicle. The camera system consists of a VG6640 high performance 1.3 megapixel CMOS image sensor and a versatile STV0991 system on chip with instant image signal processing. The high dynamic range image sensor offers supreme image quality regardless of lighting conditions thanks to its best in class front side illuminated technology with 3.75 micron pixel size. The STV0991 companion chip is a self-contained image processor that does not require external memory for its operation including for predictive H.264 video encoding. ST's state of the art algorithms for image sharpening, global tone mapping and adaptive color matrices improve on the fly HDR processing. The image processor embeds a graphics renderer that compensates for perceptual distortion produced when using a wide angle lens such as a fisheye lens. Thanks to its HDR pixel architecture, the VG6640 sensor offers an outstanding dynamic range of 132 decibels, the highest on the market. The sensor proposes various HDR configuration modes that makes this camera perfect for any tricky lighting scenes. Based on three integration times, the sensor will not miss any detail of the scene. The 3.75 micron pixel size affords the highest low light performance and the best in class signal to noise ratio in dark environments. The sensor is able to mitigate LED flicker thanks to a dedicated configuration mode. It also provides a dual conversion gain function that is particularly suitable for improving sensitivity in low light conditions. That's why the minimum dialumination is only one mililux. All this top-level performance is achievable thanks to an advanced HDR image preprocessor. In addition to performing the functions of a high-quality HDR camera, the STV0991 provides provisions for locally implementing higher-level functionalities and relieves the car's ECU from tasks that need camera-specific information. A programmable layer-specific perspective transformation enables composing the final frame with graphical elements perfectly aligned with the underlying rectified image with hardware support for scaling and rotation with bicubic resampling, alpha blending, and chroma keying. A versatile and energy-efficient lens distortion compensation block enables the use of heavily non-rectilinear lenses, such as fisheye lenses, with fully rectified fields of view close to the 180-degree limit or, for even wider, partially rectified ones. By leveraging the same set of perspective transformations, the virtualization of the camera position and its orientation is easily and effectively implemented, as, for example, in the classical tow hitch view. These advanced graphics functions are implemented without the need for full-frame buffers in order to keep latency and system cost extremely low. A dedicated image reconstruction path with specific image scaling and color space transformation support feeds two hardware-accelerated video analytics functions. The edge detection and optical flow blocks provide energy-efficient pixel-processing horsepower that enables local execution of lightweight software-implemented ADAS algorithms, such as cross-traffic alert and execution on the ECU side at largely reduced computational cost of more complex ones, such as structure from motion and objects classification. Structure from motion exploits the optical flow to compute the camera's trajectory over time and reconstruct a sparse 3D environment by means of a 3D point cloud, the colored point in the video, similarly to a LiDAR, but using a fully vision-based approach. Video analytics enable the ECU to detect pedestrians and cars by implementing an efficient object detection algorithm for when the car is stopped and the objects are moving. The embedded video analytics function also allows basic functions to be embedded directly in the STV0991's ARM Cortex R4 core without the need for external application processors. Cross-traffic alerts, known as CTAs, are an example. The HDR sensor is an ultra-low-power device, consuming less than 400 mW in streaming mode. An embedded monochrome and Bayer data preprocessor integrates a wide range of image enhancement functions. The sensor also offers numerous safety integrity features and is compliant with automotive AEC Q100 Grade II specifications. The STV0991 system-on-chip is a self-contained solution specifically designed to perform on-the-fly image processing without having to store a full-frame buffer. Both motion JPEG and H264 encoding do not require external DRAM. The embedded optical flow generator and Edge data extractor are fully implemented in hardware and consume extremely low power at full-frame operation. The versatile solution is complemented by a wide range of external interfaces for generic peripherals, including network support for both Ethernet and LVDS connectivity. To ensure greater comfort and increased safety, as well as comply with stringent regulations in the automotive market, ST's smart camera system is the perfect solution for rear view, surround view and virtual mirror replacement solutions, as well as Ethernet and other in-cabin applications. Thank you for your attention. For more information, please visit www.st.com.