A Brief Analysis on the Development Trend of Automotive Surround View Technology

 When we play racing games on the computer, the player's perspective is usually above or behind the car, so we can see the car's surroundings. This perspective can help us understand the traffic situation in real time and drive the car better. Unfortunately, we don't have such a perspective when driving a car in real life. However, this situation is gradually changing with the introduction of car surround view systems (also known as "bird's eye view" or "surround view"). Surround view is an advanced driver assistance system (ADAS) technology that can show the driver a bird's-eye 360° panoramic camera view of the car and its surroundings in real time to ensure driving safety in parking or other low-speed driving situations. In real life, there are no cameras installed on the top of the car. The bird's-eye view seen by the driver is actually a virtual view synthesized by 4 to 6 fisheye cameras installed around the car body, as shown in the figure below.

 随着环视系统被越来越多的应用在汽车上，我们从中看到了几个关键趋势。第一，视觉质量正逐步提高。早期的环视系统采用的是低分辨率摄像机，并且从未尝试应用无缝拼接。不同摄像机所拍摄的画面在整合时都会在边界用黑条替代。而更先进的系统可以实现视图的无缝地拼，让鸟瞰视图更加的真实和精确。上方所展示的示例来自基于德州仪器(TI)TDA2x的环视系统，该系统就采用了无缝拼接。实现无缝拼接需要先进的几何和光学比对算法，而TI的数字信号处理器 (DSP)就能够高效的满足此类需求。通过在环视系统中应用宽动态范围(WDR)成像算法，视觉质量还能够进一步提高。在驾驶汽车的过程中我们遇到的情况错综复杂，有时车身的不同部位会出现显著的亮度差异，例如当汽车从车库中驶出时。在种情况下，WDR成像能够帮助创建一个环视视图输出，让黑暗和明亮的区域都能够清晰可见。TI TDA3x处理器的变体拥有WDR成像能力，可应用于环视系统。

 Another potential improvement in visual quality is video noise filtering. When using a surround view system at night, the camera output often has significant noise due to low light levels. Applying advanced video noise filtering algorithms can reduce noise and improve the quality of the surround view. The TDA3x processor is capable of providing high-quality video noise filtering.

 Surround view systems typically only include a top-down bird's-eye camera view. By adding additional features, the camera perspective can be changed dynamically, so the driver can see the car from a variety of different directions and choose the perspective they want. This is made possible by GPU-based rendering. The TDA2x system-on-chip (SoC) features the SGX544 graphics engine, which can be used to implement 3D surround view with dynamic camera perspectives.

 There are other improvements to the advanced surround view system, such as improved rendering of objects located close to the car. Current surround view systems stretch such objects when rendering them. In some cases, artifacts are created when these objects are close to stitching boundaries. More advanced algorithms will reduce these distortions in the future and make the output video look more realistic.

 In addition to the visual quality improvement trends discussed above, there is another interesting trend, which is the addition of intelligent algorithms to surround view to detect certain specific events and alert the driver in emergency situations. Pedestrians and other objects around the car may cause collisions, so it is very important for the driver to know the information around the car. The visual algorithms used to analyze the surround video can detect these objects and repeatedly issue warning messages on the surround display to ensure the driver's attention. These algorithms can also enable additional sensors (such as ultrasonic) to be integrated with the surround view camera system to achieve higher robustness. In the future, we can even expect to use the surround view system to achieve autonomous driving at low speeds.