Home>>Products>>Software>>Libraries>>Image Processing Library for 'C6000
The HUNT ENGINEERING
Image Processing Library for 'C6000
Image processing applications can be built quickly using this algorithm library. Every function is heavily optimized for use on the C6x (C6000 or TMS320C6000) DSP platform. Complex algorithms can be built up using "building block" functions to produce powerful and efficient systems. Key features include:
- Extensive Range of Functions
- High performance optimized code
- Support for 8/16/32-bit images
- Concurrent DMA and CPU
- Interruptible for real-world systems
- Extendible for additional functionality
Logical Functions
Logical operators support basic pixel manipulation. Often these are used to apply
pixel masks to an image, clear areas, or set control bits within an image; they can also
be used to reduce the colour depth of the image, or provide quantization. Operators
available include:
Logical Operators | |
AND | AND two images |
ANDK | AND image with constant |
LSHK | Left-shift image |
RSHK | Right-shift image |
NOT | NOT of image |
OR | OR two images |
ORK | OR image with constant |
XOR | XOR two images |
XORK | XOR image with constant |
FILLK | Fill image with constant |
FILLRAMP | Fill image with ramp |
Arithmetic Functions
Arithmetic operators support pixel maths, allowing images to be contrast stretched,
combined with other images and so on. These operators typically vary the intensity of the
pixels in the image; for example, adding a constant to all pixels lightens the image.
Operators available include:
Arithmetic Operators | |
ABS | Absolute value of image |
ADD | Add two images |
MPY | Multiply two images |
MPYK | Multiply image by constant |
MPYKSCALE | Multiply image by constant, shift result |
MPYSCALE | Multiply two images and shift result |
SQUARE | Square the pixels in an image |
SUB | Subtract one image from another |
SUBK | Subtract a constant from the image |
Convolution Functions
Convolution can be used to implement many different operations, particularly spatial
filters and feature detectors. In HEL_ILib, it is used with standard kernel configurations
for filtering or edge detecting images. Custom kernels may also be used.
The library supports kernels with odd-numbered dimensions horizontally plus arbitrary
dimensions vertically. This allows use of all common filter kernels, such as 3x3, plus
non-square kernels like 3x7 or 9x1.