overview
In IPC, the sensor is generally used to collect the original image, and then the image is sent to the encoding chip for real-time hard coding, and then the video frame in h264 or h265 encoding format is obtained. In the Ingenic chip solution, the Sensor driver depends on the kernel and ISP driver, and the video encoding driver avpu only depends on the kernel. In addition, different Sensors and lenses also require different effect parameter configurations. The directory of the configuration file is: /etc/sensor, and the file name is: [sensor name]-[chip name].bin, for example: ov9732-t31.bin. Without this effect file, image colors, etc. may not be normal. In actual products, Sensor effect files usually need to be iteratively updated with the firmware version. Therefore, the /etc/sensor directory needs to have read and write permissions.
demand analysis
The purpose of encapsulating the video capture function is to facilitate application layer calls. So, which interfaces do we need to encapsulate? To clarify this issue, we need to think about how video is used in security surveillance scenarios.
First, the types of video streams include main stream, sub stream, and third stream. Different types of streams have different uses. The main stream is generally used for video storage, and it is necessary to ensure that the video picture is clear. Therefore, the resolution, frame rate, and bit rate will be set higher. The sub-stream is generally used for real-time video preview. It needs to ensure smooth video images and moderate network bandwidth. Therefore, the resolution, frame rate and bit rate will be set to be small. The third bit stream is generally used in other scenarios such as alarming and browsing on mobile phones. The requirements for video images are not high, and the resolution, frame rate, and bit rate are usually set to be smaller.
Secondly, various information is generally superimposed on the video screen, such as: current time, device name, preset Logo, custom picture, etc. In scenarios such as intelligent analysis, various geometric figures such as rectangular frames and polygonal frames will be superimposed on the video screen to represent detected human figures, faces,