gistfile1.py

# -*- coding: utf-8 -*-
# <nbformat>3.0</nbformat>

# <codecell>

from random import random
from SimpleCV import Image, Color
from scipy import stats


# <codecell>

 ### This is the Gaussian data smoothing function I wrote ###  
def smooth(list,degree=5):  
    window=degree*2-1  
    weight=numpy.array([1.0]*window)  
    weightGauss=[]  
    for i in range(window):  
        i=i-degree+1  
        frac=i/float(window)    
        gauss=1/(numpy.exp((4*(frac))**2))    
        weightGauss.append(gauss) 
    weight=numpy.array(weightGauss)*weight   
    smoothed=[0.0]*(len(list)-window)  

    for i in range(len(smoothed)):  
        smoothed[i]=sum(numpy.array(list[i:i+window])*weight)/sum(weight)  

    return np.array(smoothed)  

# <codecell>

def _getValues(sample):
    locations = []
    # fake the first derivative on a binary image
    d = np.where(sample[1:]-sample[:-1] < 10)[0]
    if( len(d) > 0 ):
        # count the distance between spaces
        locations = d[1:]-d[:-1]
    return locations

# <codecell>

id = 6
f = Frame.objects.order_by("-capturetime")[id]
img = f.image
self = f.features[0].feature

top = self.top[0][1]
bottom = self.lbs_line[0][1]
shaft_height = self.bottom[0][1]-bottom
head_height = bottom-top

#top = top + 0.2*head_height # move down the top this will fail with the tall bolts
#bottom = bottom #+ 0.05*shaft_height
head_left=self.lbs_left[0][0]
head_width=np.abs(self.lbs_line[0][0]-self.lbs_line[1][0])
b =    img.findBlobsFromMask(img.threshold(20).dilate(3))
dest = Image((img.width,img.height))
dest = dest.blit(b[-1].blobImage(),mask=b[-1].blobMask(),pos=b[-1].topLeftCorner())
dest = dest.applyLayers()
# move ever so slightly down the shank
span = ((bottom-top)/2.0)
spanw = span*.9 # try to avoid the fillets
temp = dest.crop(head_left,top+span,head_width,spanw)                
grain_region = temp

lines = []
meanlines = []
alllines = []
clip = 5
render = True
#oimg = grain_region.equalize().binarize(blocksize=11)
oimg = grain_region.equalize()#.binarize(blocksize=11)
temp = oimg.getGrayNumpy() # get the image -- only works on binary images
for i in range(0,oimg.height):
    result = _getValues(temp[:,i])
    if( len(result) > clip*2 ):
        result = result[clip:-1*clip]
        #lines.append(np.std(result))
        lines.append(np.mean(result))
        
        alllines.append( result ) # for each line get the length of the horizontal lines

degree = 4
window = 40

smlines = smooth(lines, degree)
curve = smooth(lines, window)

#plot(lines[degree:-degree])

oimg.dl().line((0,start), (oimg.width, start), color = Color.GREEN)
oimg.dl().line((0,end), (oimg.width, end), color = Color.RED)
oimg.save(display)


minavg = stats.scoreatpercentile(curve[:len(curve)/2], 10)
maxavg = stats.scoreatpercentile(smlines, 95)
end = np.where(smlines > maxavg)[0][0]
start = np.where(smlines[:end] < minavg)[0][-1]
while smlines[start] > smlines[start-1]:
    start -= 1

start += degree
while smlines[end] < smlines[end+1]:
    end += 1
    
end = end + degree

p = plot(smlines)
axvline(x = start - degree)
axvline(x = end - degree)

start, end

# <codecell>

plot(curve)

# <codecell>

p

# <codecell>


# <codecell>