我提取了图像并构建了两个字典，其中包含文件路径作为键，列表包含图像摘要数据作为列表。
举一个例子，我们假设我们有字典a = {}，最终看起来像{'fileName' : [ 'x=1,2,3','y=1,2,3','z=1,2,3'... ]}
以下是我的实际字典的样子：
{'/Users/cannon/AmazonBasics-Wireless-Remote-Control-for-Nikon-P700030004040x50500051006070700070s80-and90igital-SLR-Cameras.jpg': [
'fcolor=2,3,0,80,125,15,5,0,0,0,0,0,0,0,0,0,0,0,0,0,2,4,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,4,3,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0', 'edge=1,224,3,98,48,4,66,72,70,32,4,34,16,14,161,9,68,128,7', 'texture=1,173,20,9,5,3,3,4,34'],
'/Users/cannon/Sony-50mm-f-1.4-Lens-for-Sony-Alphaigital-SLR-Camera.jpg': [
'fcolor=2,111,1,18,100,1,0,0,0,0,0,0,0,0,0,1,7,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,2,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0', 'edge=1,252,35,64,99,236,127,0,17,0,1,16,139,241,24,128,2,132,31', 'texture=1,92,52,25,17,14,9,7,34']}
我编写了一个函数，它接受了我上面提到的两个字典，并且对于一个字典中的每个键，迭代第二个字典中的每个键并计算余弦值。
字典有三种类型的值转储到列表中，fcolor，edge和texture。我的函数为两个字典中的每个键计算列表中每个值的点积。
问题： 我想要做的是每个关键的Ka字典A我想要绘制字典B，Kb的所有值。
这是我目前的代码：
def cosignSimilarity(aImageVectDict, eImageVectDict):
#cx = lambda a, b : round(np.inner(a, b)/(LA.norm(a)*LA.norm(b)), 3)
fo = open('/Users/AbovePointNine.txt','w')
f1 = open('/Users/AbovePoint8.txt','w')
pltfColor = []
pltEdgeColor = []
pltTexture = []
#converting the string of vectors into numpy arrays and calculate cosine similarity
for (fn1, lst1), (fn2, lst2) in product(aImageVectDict.iteritems(), eImageVectDict.iteritems()):
print fn1
print fn2
fcolorCosine = None
edgeColorCosine = None
textureColorCosine = None
for vector1, vector2 in zip(lst1, lst2):
cx = None
cx = lambda a, b : round(np.inner(a, b)/(LA.norm(a)*LA.norm(b)), 3)
v1fColor = vector1.split('=')[1]
debug1 = v1fColor
v2fColor = vector2.split('=')[1]
debug2 = v2fColor
if (vector1.split('=')[0] == 'fcolor' and vector2.split('=')[0] == 'fcolor'):
v1fColor = np.array(map(int,v1fColor.split(',')))
v2fColor = np.array(map(int,v2fColor.split(',')))
fcolorCosine = cx(v1fColor, v2fColor)
print fcolorCosine
pltfColor.append(fcolorCosine)
v1Edge = vector1.split('=')[1]
v2Edge = vector2.split('=')[1]
if (vector1.split('=')[0] == 'edge' and vector2.split('=')[0] == 'edge'):
v1Edge = np.array(map(int,v1Edge.split(',')))
v2Edge = np.array(map(int,v2Edge.split(',')))
edgeColorCosine = cx(v1Edge, v2Edge)
print edgeColorCosine
pltEdgeColor.append(edgeColorCosine)
v1Texture = vector1.split('=')[1]
v2Texture = vector2.split('=')[1]
if (vector1.split('=')[0] == 'texture' and vector2.split('=')[0] == 'texture'):
v1Texture = np.array(map(int,v1Texture.split(',')))
v2Texture = np.array(map(int,v2Texture.split(',')))
textureColorCosine = cx(v1Texture, v2Texture)
print textureColorCosine
pltTexture.append(textureColorCosine)
if (fcolorCosine > 0.9 and edgeColorCosine > 0.9 and textureColorCosine > 0.9):
fo.write(fn1)
fo.write('\n')
fo.write(fn2)
fo.write('\n')
fo.write(str(fcolorCosine))
fo.write('\n')
fo.write(str(edgeColorCosine))
fo.write('\n')
fo.write(str(textureColorCosine))
if (fcolorCosine > 0.8 and edgeColorCosine > 0.8 and textureColorCosine > 0.8):
f1.write(fn1)
f1.write('\n')
f1.write(fn2)
f1.write('\n')
f1.write(str(fcolorCosine))
f1.write('\n')
f1.write(str(edgeColorCosine))
f1.write('\n')
f1.write(str(textureColorCosine))
fo.close()
f1.close()
#Plotting
plt.plot(pltfColor, 'r.')
#plt.plot(pltEdgeColor, 'b.')
#plt.plot(pltTexture, 'g.')
plt.ylabel('cosine values')
#plt.yscale('log')
plt.show()
fo.close()
再次确认我真正想要的是对于词典A中的每个键，我想找出字典B中的所有结果