[R-sig-Geo] problems extracting raster values from x,y positions

[Benjamin Mack]

Hello all,
I'm getting crazy with a really basic problem but can't figure out what 
I am doing wrong. I found a way (see below) to solve the problem but it 
is everything but not elegant.
I have an raster img with 72 layers which I read from an envi file.

 > img <- stack(fname)

Now I want to extract some pixel values using an SpatialPointsDataFrame 
img_vals <- extract(img, ref)
The problem is, that the output tells my that the values in the 72 bands 
are all equal - which is not true!

 > img_vals
  [1,]         0.034341563           0.034341563 ...
  [2,]         0.014481238           0.014481238 ...
  [...,]

To show you the strange behaviour look at the following.

Trying to do the following (which I suppose to be somehow equal than 
using extract but with the pixel row/col -locations) I get the same 
problem (equal values in all bands):

 >img[1,1]
[1,]         0.008778242           0.008778242  ...

BUT when I do it in a different way I get the truth (controlled in ENVI) 
with differnet values in each layer.

 > getValues(img at layers[[1]])[1]
[1] 0.008778242
 > getValues(img at layers[[2]])[1]
[1] 0.006187854
 > getValues(img at layers[[3]])[1]
[1] 0.01333527

Has anybody an idea what is wrong here and how I have to use the extract 
command correctly? Any suggestions are highly appreciated!

Thanks in advance!

---------------------------------------------------
My (ugly) workaround:

img <- stack(fname)
ci <- rasterize(ref,img,'ID')
ind_sampling <- which(is.na(getValues(ci))==F)

vals <- c()
for (ii in 1:dim(img)[3]) {
   vals <- rbind(vals, getValues(img at layers[[ii]])[ind_sampling]) }