def featureSearch(queryImage,feats):
# 提取查询图像的特征，计算 simlarity 评分和排序
queryVec = extract_feat (queryImage)
scores = np.dot (queryVec, feats.T) # 计算点积（内积）,计算图像得分
# 矩阵乘法并把（纵列）向量当作n×1 矩阵，点积还可以写为：a·b=a^T*b。
# 点积越大，说明向量夹角越小。点积等于1，则向量为同向，向量夹角0度。
rank_ID = np.argsort (scores)[::-1] # 排序,倒序，大到小
rank_score = scores[rank_ID] # 计算评分
# print("scores",scores,type(scores))
# print ("rank_ID",rank_ID,type(rank_ID))
# print ("rank_score",rank_score,type(rank_score))
return rank_ID,rank_score
def etlFeature(post, img_list, h5filename):
# 迭代方式，提取特征值写入h5文件
bar = ProgBar(len(img_list),
monitor=True,
title="提取图片特征,Image Total:%d" % len(img_list))
for i, img_path in enumerate(img_list):
norm_feat = extract_feat(img_path)
img_name = os.path.split(img_path)[1]
names = []
names.append(img_name)
feats2 = np.array(norm_feat)
try:
wH5FileData(i + post, feats2, names, h5filename)
except:
print("Feats Write Error")
return 1
bar.update()
# print ("提取图片特征！进度: %d/%d" % ((i + 1), len (img_list)))
print(bar)
return 0
Extract features and index the images
'''
if __name__ == "__main__":
db = args["database"]
img_list = get_imlist(db)
print ("--------------------------------------------------")
print (" feature extraction starts")
print ("--------------------------------------------------")
feats = []
names = []
for i, img_path in enumerate(img_list):
norm_feat = extract_feat(img_path)
img_name = os.path.split(img_path)[1]
feats.append(norm_feat)
names.append(img_name)
print ("extracting feature from image No. %d , %d images in total" %((i+1), len(img_list)))
feats = np.array(feats)
# names = np.array(names)
# directory for storing extracted features
output = args["index"]
print( "--------------------------------------------------")
print (" writing feature extraction results ...")
print ("--------------------------------------------------")
print(output)
print(feats.shape)
imgNames = h5f['dataset_2'][:]
h5f.close()
print "--------------------------------------------------"
print " searching starts"
print "--------------------------------------------------"
# read and show query image
queryDir = args["query"]
queryImg = mpimg.imread(queryDir)
plt.title("Query Image")
plt.imshow(queryImg)
plt.show()
# extract query image's feature, compute simlarity score and sort
queryVec = extract_feat(queryDir)
scores = np.dot(queryVec, feats.T)
rank_ID = np.argsort(scores)[::-1]
rank_score = scores[rank_ID]
#print rank_ID
#print rank_score
# number of top retrieved images to show
maxres = 3
imlist = [imgNames[index] for i, index in enumerate(rank_ID[0:maxres])]
print "top %d images in order are: " % maxres, imlist
# show top #maxres retrieved result one by one
for i, im in enumerate(imlist):
image = mpimg.imread(args["result"] + "/" + im)
plt.title("search output %d" % (i + 1))
h5f.close()
print "--------------------------------------------------"
print " searching starts"
print "--------------------------------------------------"
# read and show query image
queryDir = args["query"]
queryImg = mpimg.imread(queryDir)
plt.title("Query Image")
plt.imshow(queryImg)
plt.show()
# extract query image's feature, compute simlarity score and sort
queryVec = extract_feat(queryDir)
scores = np.dot(queryVec, feats.T)
rank_ID = np.argsort(scores)[::-1]
rank_score = scores[rank_ID]
#print rank_ID
#print rank_score
# number of top retrieved images to show
maxres = 3
imlist = [imgNames[index] for i,index in enumerate(rank_ID[0:maxres])]
print "top %d images in order are: " %maxres, imlist
# show top #maxres retrieved result one by one
for i,im in enumerate(imlist):
def process_classify_1_by_2(request):
# 获取页面传来的数据
if request.method == "POST":
data = request.POST.get("image")
print(data)
data = data[1:]
label = str(request.POST.get("label"))
# print(label)
# print(data)
# 相似性查询，调用训练的相似性模型来查找相似的图片
h5f = h5py.File('quanshanshui_index6688', 'r')
feats = h5f['dataset_1'][:]
imgNames = h5f['dataset_2'][:]
h5f.close()
queryDir = data
queryVec = extract_feat(queryDir)
scores = np.dot(queryVec, feats.T)
rank_ID = np.argsort(scores)[::-1]
maxres = 20
i = 0
imlist1 = [
imgNames[index] for i, index in enumerate(rank_ID[1:maxres])
]
print("*****************************************")
print(imlist1)
images = {}
yuantu = {}
result = []
i = 0
# 查找到相似图，并计算出框在原图的位置
for j in imlist1:
j = str(j)
j = j[2:-5]
print(j)
obj = models.Psection.objects.filter(sectionname=str(j))
if obj.exists():
# print("**********")
str1 = "/" + str(obj[0].section_url)
label2 = str(obj[0].label)
# print(str1)
# print(label2)
# print(label)
if label == label2:
images[i] = str1
obj2 = models.Painting.objects.filter(id=obj[0].pid)
gg = "/media/" + str(obj2[0].picture_url)
filename = obj2[0].filename
print("filenme", filename)
border_list = []
yu_width = obj2[0].width
yu_height = obj2[0].height
xmin = obj[0].xmin
ymin = obj[0].ymin
xmax = obj[0].xmax
ymax = obj[0].ymax
border_width = xmax - xmin
border_height = ymax - ymin
# print("______________")
# print(obj2[0].id)
# print(filename)
# print("图的宽度 %s"%yu_width)
# print("图的高度 %s"%yu_height)
# print("$$$$$$$$$")
# print("xmin %s"%xmin)
# print("xmax %s"%xmax)
# print("边框的宽度%s"%border_width)
# print("ymin %s"%ymin)
# print("ymax %s" %ymax)
# print("边框的高度%s"%border_height)
# print("$$$$$$$$$")
h_w_rate = 1
w_h_rate = 1
h_rate = 1
w_rate = 1
if yu_width > yu_height:
w_h_rate = yu_width / yu_height
print("w_h_rate %s" % w_h_rate)
if yu_width > 590:
w_rate = yu_width / 590
h_rate = yu_height / (590 / w_h_rate)
# print("yu_width > 590:")
# print("w_rate%s"%w_rate)
# print("h_rate%s"%h_rate)
elif yu_height > 700:
# print("yu_height > 700:")
# print("w_rate %s" %w_rate)
# print("h_rate%s" %h_rate)
h_rate = yu_height / 700
w_rate = yu_width / (700 / w_h_rate)
else:
h_w_rate = yu_height / yu_width
print("h_w_rate %s" % h_w_rate)
if yu_height > 700:
h_rate = yu_height / 700
w_rate = yu_width / (700 / h_w_rate)
# print("yu_height > 700:")
# print("w_rate%s" % w_rate)
# print("h_rate%s" % h_rate)
elif yu_width > 590:
w_rate = yu_width / 590
h_rate = yu_height / (590 / h_w_rate)
# print("yu_width > 590:")
# print("w_rate%s" % w_rate)
# print("h_rate%s" % h_rate)
# print(h_rate)
# print(w_rate)
border_height = border_height / h_rate
border_width = border_width / w_rate
xmin = xmin / w_rate
ymin = ymin / h_rate
# print("xmin %s" % xmin)
# print("ymin %s" % ymin)
# print("边框的宽度%s" % border_width)
# print("边框的高度%s" % border_height)
# print("______________")
border_list.append(xmin)
border_list.append(ymin + 55)
border_list.append(border_width)
border_list.append(border_height)
border_list.append(filename)
border_list.append(gg)
yuantu[i] = border_list
i += 1
# images[0]='/media/tikuan/tikuan13001203.jpg'
# images[1]='/media/yinzhang/yinzhang1513432.jpg'
# images[2]='/media/fantou/fantou12005074.jpg'
print("相似查询")
print(images)
print(yuantu)
print("相似查询")
content = render_to_string("divtwo.html", {
"image123": images,
"yuantu": yuantu
})
result.append(content)
return HttpResponse("".join(result))
Extract features and index the images
'''
if __name__ == "__main__":
db = args["database"]
img_list = get_imlist(db)
print "--------------------------------------------------"
print " feature extraction starts"
print "--------------------------------------------------"
feats = []
names = []
for i, img_path in enumerate(img_list):
norm_feat = extract_feat(img_path)
img_name = os.path.split(img_path)[1]
feats.append(norm_feat)
names.append(img_name)
print "extracting feature from image No. %d , %d images in total" %((i+1), len(img_list))
feats = np.array(feats)
# directory for storing extracted features
output = args["index"]
print "--------------------------------------------------"
print " writing feature extraction results ..."
print "--------------------------------------------------"
h5f = h5py.File(output, 'w')
h5f.create_dataset('dataset_1', data = feats)