MJPEG Encoder

The MJPEG encoder provides high-quality still and video compression with a wide quality range, and it supports an industry-standard algorithm.

To learn more about using the MJPEG encoder, see:

• File Format
• Algorithm Overview
• Encoding a JPEG Frame
• Tips for Using the UMSJPEGEncoder

For introductory information, see Programming with Video Codec Objects.

File Format

The JPEG syntax includes all the information that a decoder needs to decompress a frame, independently of any file format that can happen to contain the compressed frame.

However, JPEG does not specify all the information that might be needed to format and display a decompressed frame. For example, it does not include fields to specify the color space of the compressed data or the pixel dimensions of the decompressed frame; rather, the AVI file format supplies this extra information. The JPEG objects can be used by themselves to support raw JPEG, or they can be used with the UMSAVIReadWrite object to support JPEG-in-AVI.

Note: At the time Ultimedia Services was under development, the definition for JPEG and MJPEG in AVI was in draft form. This draft did not clearly define a standard location for the Huffman tables. This implementation is based on information available at the time and can be modified in future releases to comply with a final standard. This implementation includes the Huffman tables (the X'FF' DHT segment) in each frame of an AVI sequence rather than using an abbreviated JPEG format. Since this is suspected to be different from the MJPEG standard, the biCompression field of the BITMAPINFOHEADER is coded "mJPG" rather than "MJPG".

Algorithm Overview

The JPEG standard includes a baseline algorithm, extended algorithms, and an independent lossless algorithm. (The lossless algorithm is unrelated to the other algorithms.) The extended algorithms build on the baseline algorithm to add extra capabilities, directed primarily at the transmission of still images over low-bandwidth channels. They allow for the refinement of a decompressed image over time. Finally, JPEG provides for the use of either Huffman or arithmetic coding as the last stage of the compression.

Note: The most popular algorithm, by far, is the baseline algorithm with Huffman encoding. This is the only algorithm that Ultimedia Services supports. In the remainder of this section, JPEG refers to this algorithm.

JPEG supports arbitrary image sizes. Also, JPEG is a symmetric algorithm--the times to encode and decode a given image are similar.

Although JPEG was designed for still images, it is often used to compress video as well. While codecs that use interframe correlations can achieve better compression, JPEG offers the advantages of fast compression and easier editing than if delta frames are used.

The JPEG standard treats the different components of a frame separately and includes no color-conversion. Also, it allows for subsampling of the different components relative to the full size of the frame.

Despite this flexibility, the most common color space for JPEG-compressed images is the YCbCr (YUV) color space. For this color space, the component subsampling is limited to the chrominance (Cb and Cr or U and V) components and the subsampling factors are small. The JPEG encoder and decoder objects support subsampling of the chrominance components by 2 in either or both directions.

The most common application of JPEG is to compress RGB images after they have been converted to the YCbCr color space. Therefore, the JPEG encoder and decoder objects integrate color conversion with the JPEG codec.

Encoding JPEG Frames

To encode JPEG frames, do the following:

1. Create a UMSJPEGEncoder object.
2. If necessary, create a file reader object.
3. Call the encoder's set_image_format and set_image_size methods. To encode a subimage of each frame, optionally call set_sub_image_size after set_image_size.
4. Set the quality of the compression with the encoder's set_quality_factor method.
5. Set the chrominance subsampling with the encoder's set_subsample method.
6. Allocate memory for the buffers for the input and compressed frames.
7. Get the input frame and call the encoder's compress_frame_seq method to compress the frame.

Repeat step 7 for more frames.

When the video is finished, free all buffers. When the objects are no longer needed, destroy them with _somFree.

Tips for Using the UMSJPEGEncoder

The following tips provide important information for using the UMSJPEGEncoder:

• The encoder supports the image formats RGB24, Mono8, and a separate-component RGB format. The later format consists of three separate 8-bit/pixel raster-scan components.
• The quality factor can vary from 0 to 100.
• This object does not have a get_max_buffer_size method. The complexity of the JPEG syntax makes it difficult to compute an upper limit on the size of a compressed frame. If the output buffer size is equal to the size of an input frame, that should be enough. For color input, the output buffer can be one-fourth the size of an input frame for the RGB24 format and one-third the size for the RGB format.
• There is a get_comp_frame_size that can be used after a frame is compressed.
• The encoder supports up to 2:1 subsampling of the chrominance components in either direction.
• The default settings for the encoder follow:

  chrominance subsampling		none
  frame dimensions		0 x 0
  subimage dimensions		0 x 0
  subimage coordinates		(0, 0)

For introductory information, see Programming with Video Codec Objects.