Image acquisition method, device and optical path conversion element

Abstract

This manual discloses a method, device and light path transformation element for image collection.The camera in this description embodiment includes sensors, light path transformation elements, and at least two lenses. Among them, each lens of the camera is configured with different colors of filter; the transformed element of the light path is used for each lens, so that the lens is shot from the lens into the lens into the lens.The light path of light is transformed to projected the light after the transition of the transformed light to the sensor on the sensor corresponding to the color sensor component of the color of the filter configuration configured with the lens.And when collecting images, at least two lenses of the camera are controlled according to the preset exposure parameter to expose at the same time, and the image is generated; for each lens, the color corresponding to the color corresponding to the filter in the image to the shot configuration isThe weight is obtained based on the color component.In this way, the hardware cost can be reduced while obtaining multiple images simultaneously, and the data transmission efficiency is improved by reducing the total data of the collected image.

Description

technical field [0001] This specification relates to the field of unmanned driving technology, and in particular to an image acquisition method, device and optical path conversion element. Background technique [0002] With the rapid development of intelligent driving and artificial intelligence, machine vision, as an important technology for unmanned vehicles to perceive the environment, has also received extensive attention. [0003] In the prior art, when an unmanned vehicle perceives the surrounding environment through machine vision technology, it usually deploys multiple cameras at different positions on the unmanned vehicle to obtain image information from multiple directions around the unmanned vehicle, and then the obtained image The information is transmitted to a computer for analysis. When the unmanned vehicle is in the process of driving, the unmanned vehicle needs to sense the surrounding environment in real time like a human, and make corresponding operations...

AI Technical Summary

Problems solved by technology

However, in the prior art, because the line width and line length of the I / O (input / output, input / output) transmission port between the camera and the computer are limited by engineering design, the I / O transmission port between the camera and the computer There is also an upper limit to the total amount of data transmitted in real time

[0005] In this way, when the unmanned vehicle is in the process of driving, when the camera transmits the collected image data to the computer, there may be a problem that the data volume of the image data to be transmitted is too large, resulting in low transmission efficiency, which in turn leads to the failure of the unmanned vehicle. The problem of being unable to respond to environmental changes in real time

[0006] In addition, because the working environment of the camera deployed on the unmanned vehicle is complex and changeable, and it is often used for high-speed shooting, and at the same time requires a long service life, the requirements for the camera in unmanned driving are relatively high, which makes unmanned The price of cameras used in human driving is often relatively high. When multiple cameras are deployed to achieve multi-directional visual image acquisition, the camera input cost will be too high

[0007] Furthermore, when using multiple cameras to obtain visual images in multiple directions, it is necessary to pay a high cost to realize the height synchronization of multiple cameras on the unmanned vehicle; therefore, in the prior art, a certain degree of Synchronization accuracy to maintain low production costs. In this way, there is a certain difference between the times when images are collected by multiple lenses, resulting in multiple images collected reflecting the environment of multiple directions of the unmanned vehicle at different times, and based on these images The results of analysis and processing cannot accurately reflect the environmental changes around the unmanned vehicle, which eventually leads to the problem that the unmanned vehicle cannot accurately respond to environmental changes in real time.

[0008] To sum up, in the existing unmanned driving solutions, when acquiring visual images from multiple directions, there are high costs, it is impossible to achieve a high degree of camera synchronization while maintaining low cost, and there may be a large amount of image data to be transmitted. The problem of low transmission efficiency caused by too large

Images