Howell released the world’s first product level CIS/EVS integrated vision chip OV60B10
2022-11-26
[Guide] With the increasing diversification of various terminal application scenarios, the demand for the technical upgrading of visual sensors is not only to improve some basic parameters such as pixels, light input, white balance, but also to make the products more intelligent and comprehensive to cope with the complex and changing external environment.
For example, in head display devices such as AR/VR, eye tracking technology has always been an important part of terminal upgrading. On the one hand, due to the different signs of each user, the conventional eye tracking technology cannot achieve large-scale adaptation; On the other hand, in such scenes as gaze rendering, strict requirements are put forward for low delay, accuracy and predictability of sensor eye tracking function.
In addition, in the field of smart phones, more and more users hope to complete professional photography, such as light photography, dark indoor environment photography, outdoor static photography, and even real-time photography of objects in ultra-high speed motion through mobile phones. Such applications not only require sensors to provide sufficient dynamic range, so as to achieve more comprehensive and detailed scene coverage; It is also required that the sensor can provide clear and accurate image processing results when the subject moves at a high speed.
In addition to the above, multiple applications, including ADAS, machine vision, object tracking, etc., also need a powerful vision processor to further optimize the user experience to support their upgrading.
For the demand derived from these specific scenarios, Howell Group recently released the world's first product level CIS/EVS integrated vision chip OV60B10. Through the fusion of the two, the characteristics of the two types of sensors are integrated on a single chip.
EVS (Event Based Vision Sensors) is a new type of bio inspired vision sensor, which has the technical characteristics of event based vision, high dynamic range, high speed, low delay, no motion blur, and low data rate and low power consumption. Compared with the widely used CIS (CMOS Image Sensor), the output of the event camera is continuous in space and time, and its response speed is not limited by the traditional exposure time and frame rate. It can detect the ultra-high speed movement as fast as a bullet, while using the CIS camera requires thousands or tens of thousands of frames per second to capture it; It is not affected by unified imaging parameters such as white balance, sensitivity, exposure time, etc. It can still trigger events to obtain visual information when the image is too dark, overexposed, or changes dramatically.。
Technical highlights
1. 3D Stack Process
OV60B10 chip is manufactured by 3D Stack process. By integrating CIS, EVS, ISP/ESP three-layer wafers into the state of optimal performance and minimum volume. High density pixel level connection between wafers is realized by hybrid bonding technology. Because the detector part and the circuit part are independent, the processing technology is optimized separately and the pixel unit size is minimized.
2. Pixel level sensor fusion
The OV60B10 chip can share the focal plane through two kinds of sensors, and has multiple signal interactions, so as to match with high accuracy in time and space, work together, and output in parallel (it can also work independently through software configuration).
Through pixel level sensor fusion technology, the advantages of the two sensors are maximized, and the redundant information and complementary information in the image are fully used to generate an image that meets the needs of specific applications. In this way, the same scene or target can be described more accurately, comprehensively and reliably.
3. World leading EVS technology
OV60B10 chip adopts the in pixel timestamp and related high-speed readout technology unique to Howell Group. Compared with the existing timestamping reading mode in the industry, OV60B10 chip can greatly improve the time scale accuracy of EVS and reduce the jitter of the time scale. OV60B10 has also innovated a series of technologies such as noise control, multi-channel readout, event coding and dynamic bandwidth.
The more frames per second, the smoother the motion will be displayed. Through the exclusive EVS technology, OV60B10 reconstructs the original frame rate of only 120 FPS, and the frame rate can reach 10,000 FPS. The terminal can use this sensor to capture high-speed images with less data, less computing and lower hardware costs.
4. Excellent CIS imaging performance
OV60B10 chip also uses the industry-leading CIS platform, which has the advantages of high resolution (15 million pixels) and large pixel (2.2um).
To sum up, the OV60B10 chip enables the event camera with the characteristics of efficiently capturing scene changes, low delay, low data volume, etc., and the CIS with high resolution and large pixels to display their strengths. It is suitable for ultra-high speed image reconstruction, high dynamic range imaging, ADAS, intelligent cockpit, eye tracking, object tracking, SLAM and other scenes. It allows users to flexibly respond to the shooting needs of mobile phones, cars, AR/VR and other different fields in the way of "combination boxing".
From the perspective of the terminal, this visual sensor with outstanding performance can also help it to conduct deeper product development and mining.
It is well known in the industry that the rise of the concept of "meta universe" has once again attracted great attention to AR/VR devices. However, for a long time, the development of such devices has been constrained by the lack of content and interactive experience. At present, the terminal can realize high-precision and high-speed motion capture with OV60B10 ultra-high speed image reconstruction, low delay and other functions, so as to provide users with smoother, more immersive and richer game scenes. Deeper into the concept of "meta universe", it can even be found that the accuracy of motion track capture actually determines the degree of digitalization of user actions, expressions and other indicators, and the degree of digitalization also determines whether users can present their own unique image and achieve interaction in the digital world.