Effective Reduction of Memory Storage and Cost by HEVC-based Intra-Frame Video Encoding Technology with Region-Adaptive Quantization for Car Digital Video Recorders

• Published in: 2022 4th International Conference on Computer Communication and the Internet (ICCCI) pp. 62 - 71
• Main Authors: Chen, Chih-Hsiang, Fan, Chih-Peng
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.07.2022
• Subjects: Cameras; Car Digital Video Recorders; HEVC; Image coding; Integrated circuits; Memory management; Memory Storage; Quantization (signal); Region-Adaptive Quantization; Surveillance; Transportation
• DOI: 10.1109/ICCCI55554.2022.9850269

Dashcams are critical devices for ensuring transportation safety. Specifically, they provide onsite records for transportation incidents, which allows investigators to analyze the causes of the incidents. Therefore, dashcams are commonly used in public transport, automobiles, and motorcycles. Current dashcams can achieve 1080p and 4K resolution. In addition to high-resolution charge-coupled devices and components, dashcams require a powerful video encoder integrated circuit to encode images. The results obtained after image encoding are stored in memory cards or other digital memory components. The difference between commercial dashcams and general surveillance cameras lies in their memory components. The memory components of surveillance cameras are typically cheap and easily obtainable Secure Digital (SD) cards and flash ICs, whereas dashcams require small and cheap memory components. In contrast to general surveillance equipment, dashcams do not contain a large-scale hard drive matrix and do not have access to a large server for maintaining records. Thus, dashcams must achieve an effective compression efficiency without causing distortion. Most SD cards on the market have a storage size of 8-64 GB. Advances must be achieved in image compression technology so that the image compression ratio can be increased to reduce the video size without causing problems in subsequent image reading and identification. In this study, the file compression ratio achieved with the developed technology was 1.7-2.1 times higher than that achieved without the application of this technology. Moreover, the storage space saving ratio of the aforementioned technology was 42%-53%.