nalankaru / color_hmax

Hmax color

(1) Single-Opponent (SO) and Double-Opponent (DO) descriptors

SODescriptor.m / DODescriptor.m: Compute SO/DO descriptors energyRes.m: Compute energy response based on Adelson & Bergen (1985)

We provide two examples to show how color descriptors work:

1. additive color image used in Zhang et al (2012)
2. blue-sky image: a representative color image

The normalization paramaters may need to be adjusted depending on the type of images. In general, we found k = 1, sigma = 0.225 to work well for natural images.

---

(2) Color HMAX

demoRelease.m : grayscale Hmax demoSoRelease.m: SOHmax (for most cases, two orientations of SO is sufficient because of their weak orientation tuning) demoDoRelease.m: DOHmax

Dictionary of S2 units: randomly extracted -- 250 patches of 4 patch sizes (1000 patches in total)

We provide three types of dictionaries: dict_250_patches_4_sizes : grayscale dictSo_250_patches_4_sizes: SO dictDo_250_patches_4_sizes: DO

If you use the code, please cite: Zhang J., Barhomi Y., and Serre T. A new biologically inspired color image descriptor.In: ECCV, Florence, Italy, October 2012.

For comments or questions, please contact Jun Zhang ([e-mail]) or Thomas Serre ([e-mail]).