TI DSP库VLIB介绍

The Texas Instruments VLIB is an optimizedImage/Video Processing Functions Library for C programmers using c674x devices.It includes many C-callable, assembly-optimized, general-purpose image/videoprocessing routines.

运行这些实例的速度要比用ANSI C编写的同样代码效率高，性能可提高10倍，可以大大减少开发时间。

可供用于：

• 背景建模&相减

• 物体特征提取

• 跟踪与检测

• 低级像素处理（low-levelpixel processing）

安装之后，预编译的库有：

• vlib.lib : Main optimized vlib library
• vlib_cn.lib : C Natural Implementation of vlib library
• common.lib : Common functions required for testing kernels.

 

Integration

对于单核或者多核系统，集成时需要以下步骤：

在系统代码中调用API

        需要包含vlib.h

编译代码

        编译时，编译器需要能够搜索packages目录，以找到kernelheaders

链接VLIB

        -lvlib库所在的目录

 

 

支持的函数：

2. Exponentially-Weighted Running Mean of a Video
3. Exponentially-Weighted Running Variance of a Video
4. Uniformly-Weighted Running Mean of a Video
5. Uniformly-Weighted Running Variance of a Video
6. Statistical Background Subtraction
7. Mixture of Gaussians Background Modeling for Grayscale Video
8. Image Extraction From Background Models (8-Bit)
9. Packing and Unpacking of Binary Mask Images
10. Morphological Dilation
11. Morphological Erosion
12. Connected Components Labeling
13. Canny Edge Detection
14. Image Pyramid
15. Gaussian 5x5 Pyramid Kernel
16. Gradient 5x5 Pyramid Kernel
17. Recursive IIR Filter: Horizontal, First-Order
18. Recursive IIR Filter: Vertical, First-Order
19. Integral Image
20. Hough Transform for Lines
21. Harris Corner Score
22. Non-Maximal Suppression (mask output)
23. Non-Maximal Suppression (list output)
24. Lucas-Kanade Feature Tracking (Sparse Optical Flow)
25. Normal Flow
26. Kalman Filter
27. Nelder-Mead Simplex
28. Legendre Moments
29. Histogram
30. Bhattacharya Distance
31. L1 Distance
32. Color Space Conversions
33. SAD Based Disparity
34. Image Rescaling
35. Histogram Equalization
36. Hog Cell
37. Good Features to Track
38. L2_normalization
39. Gradients, Magnitude, and Orientation
40. Hysteresis Thresholding
41. Grayscale Morphology

参考：VLIB User's Manual (c674x)

本人从事DM8168 DSP开发，欢迎加QQ：917581001讨论