Google has launched a new open-source library jpegli, which implements an encoder and decoder for JPEG format images. The library includes additional optimizations to enhance encoding efficiency, allowing a 35% increase in compression rates for high-quality images compared to traditional JPEG codecs. In comparison to libjpeg-turbo, the jpegli library achieves a similar quality level while reducing the bitrate by 32%. At the API and ABI levels, the library is fully compatible with libjpeg62 and can be used for its seamless replacement. The library's code is written in C++ and is distributed under the BSD license.
The increase in compression levels is achieved through advanced technologies that reduce image noise and enhance quality, utilizing more efficient psychovisual modeling methods to diminish artifacts. Specifically, jpegli employs adaptive quantization heuristics used in the JPEG XL project, along with improved algorithms for selecting quantization matrices and calculating intermediate results.
The jpegli library allows encoding images with 10 or more bits per color component. The output of the new encoding algorithms is adapted for the traditional JPEG model, which only allows for 8 bits per color component. This feature ensures compatibility with existing decoders designed for 8-bit representations of color components.
Images encoded with jpegli fully comply with the JPEG standard, do not require specific decoders, and can be viewed in existing JPEG viewers and web browsers. Using its own decoder for unpacking images compressed with jpegli allows for further reduction of artifacts. The encoding speed with jpegli is comparable to that of the libjpeg-turbo and MozJPEG libraries.
Source: opennet.ru
