SelectFocusIMAGE is an image compression software tool that combines a state-of-the-art, region-of-interest-based image codec, an easy-to-use and intuitive graphical user interface, and a fault-tolerant IP-based transport mechanism. The software is versatile, easy to use, and compatible with all Windows, Unix, and Linux systems, and will operate over standard IP-based networks. This article outlines the features and capabilities of SelectFocusIMAGE, and explains its three main components.
ROI-Based Image Codec
SelectFocusIMAGE is designed for both uniform and region-of-interest-based compression. In the former, a quota of bits is applied to the image such that the degradation induced by the compression process is spread uniformly throughout the image.
In region-of-interest-based compression, more bits are used to code the targets or regions of interest, and fewer bits are used to code the background, which is typically used only for contextual purposes. This ROI-based coding methodology enables the target regions within the image to be well preserved, while dramatically reducing the number of bits required to
represent the image, thus reducing bandwidth and storage requirements.
The compression algorithm developed for SelectFocusIMAGE is a specially modified version of JPEG2000. Recall that JPEG2000 is a widely used international standard for still image compression, and is much more efficient at coding images than JPEG. JPEG2000 incorporates a wide array of features, including the ability to support region-of-interest coding as outlined above.
Unfortunately, the coding efficiency of JPEG2000 degrades substantially when coding multiple regions of interest. Because of this degraded performance, standard JPEG2000 was found unsuitable for low-bit-rate applications where multiple regions of interest were specified. This served as the impetus for the development of the SelectFocusIMAGE compression core.
View full size
Table A: SelectFocusIMAGE features and benefits
Compression performance compared
To illustrate the performance of SelectFocusIMAGE as compared to standard JPEG2000, multiple regions of interest in Figure 1 were coded using both SelectFocusIMAGE and the JPEG2000 scaling-based method, each at a bit rate of 0.004 bits/pixel (or a compression ratio of 2000:1). Note that the scaling-based method of JPEG2000 can only accommodate rectangular and elliptical ROIs, while SelectFocusIMAGE can accommodate ROIs of arbitrary shape. (Note: Click on "View full size" to clearly see differences.)
From the figure, it is apparent that the image coded using SelectFocusIMAGE is dramatically superior to the JPEG2000-compressed image both quantitatively and subjectively.
In fact, SelectFocusIMAGE outperforms JPEG2000 by 3.67 dB, which equates to the SelectFocus-compressed image requiring less than half of the channel bandwidth required for transmission of the compressed image as compared to JPEG2000.
View full size
Figure 1: Coding performance of SelectFocusIMAGE as compared to JPEG2000
|
