def com2imgs():
if request.method == "POST":
imgbase64_1 = request.form.get('imgbase64_1')
imgbase64_1 = base64.b64decode(imgbase64_1)
imgbase64_2 = request.form.get('imgbase64_2')
imgbase64_2 = base64.b64decode(imgbase64_2)
randname1 = getRandomSet(20) + '.jpg'
randname2 = getRandomSet(20) + '.jpg'
file = open(randname1, 'wb')
file.write(imgbase64_1)
file.close()
file = open(randname2, 'wb')
file.write(imgbase64_2)
file.close()
try:
img1 = cv2.imread(randname1)
img2 = cv2.imread(randname2)
hash1 = fun_Hash(img1)
hash2 = fun_Hash(img2)
score = com2hashstr(hash1, hash2)
os.remove(randname1)
os.remove(randname2)
return {'sign': 1, 'score': score}
except:
os.remove(randname1)
os.remove(randname2)
return {'sign': -1, 'score': None}
else:
return "<h1>compare img, please use post !</h1>"
def get_limit_imgpath(dstimgpath, _hash_strs, _imgpaths):
limit = 85
dstimg = cv2.imread(dstimgpath)
dstimg_2 = cv2.flip(dstimg, 1) # check overturn image on the same time
dst_hash_str = fun_Hash(dstimg)
dst_hash_str_2 = fun_Hash(dstimg_2)
out_imgpaths = []
for n in range(len(_hash_strs)):
temp_value = com2hashstr(dst_hash_str, _hash_strs[n])
temp_value_2 = com2hashstr(dst_hash_str_2, _hash_strs[n])
temp_value = max(temp_value, temp_value_2) # use max score
print('5444646464646464', temp_value)
if temp_value > limit:
out_imgpaths.append(_imgpaths[n])
return out_imgpaths
def load_img(orgimgroot):
hash_str_list = []
imgpath_list = []
orgimg_list = os.listdir(orgimgroot)
cnt = 0
cnt_error = 0
for n in orgimg_list:
imgpath = os.path.join(orgimgroot, n)
print('>>>', cnt, imgpath)
cnt = cnt + 1
try:
img = cv2.imread(imgpath)
img.shape
except:
print('read img error: ', imgpath)
cnt_error = cnt_error + 1
continue
img_hash = fun_Hash(img)
hash_str_list.append(img_hash)
# 拼接es数据库图像url，接下来仍然需要调整的 ！！！！！！！！！！
imgpath = [
"allimage", "https://filefusion.ccwb.cn/files/2020/12/images/" + n
]
imgpath_list.append(imgpath)
print('successful and error: ', cnt, cnt_error)
return hash_str_list, imgpath_list
def load_img(imgs_record_txt):
hash_str_list = []
img_from_id_list = []
# orgimg_list = os.listdir(orgimgroot)
cnt = 0
cnt_error = 0
for n_ in open(imgs_record_txt):
n = n_[:-1]
imgpath = n.split(',')[0]
print('>>>', cnt, imgpath)
cnt = cnt + 1
try:
img = cv2.imread(imgpath)
img.shape
except:
print('read img error: ', imgpath)
cnt_error = cnt_error + 1
continue
img_hash = fun_Hash(img)
hash_str_list.append(img_hash)
# 拼接es数据库图像url，接下来仍然需要调整的 ！！！！！！！！！！
img_from_id = n.split(',')[1]
img_from_id_list.append(img_from_id)
print('successful and error: ', cnt, cnt_error)
return hash_str_list, img_from_id_list # 图像hash和来源id对应
def get_limit_imgpath(dstimgpath, _hash_strs, _imgpaths, limit):
limit = int(limit)
dstimg = cv2.imread(dstimgpath)
dstimg_2 = cv2.flip(dstimg, 1) # check overturn image on the same time
dst_hash_str = fun_Hash(dstimg)
dst_hash_str_2 = fun_Hash(dstimg_2)
out_imgpaths = []
similar_scores = []
for n in range(len(_hash_strs)):
temp_value = com2hashstr(dst_hash_str, _hash_strs[n])
temp_value_2 = com2hashstr(dst_hash_str_2, _hash_strs[n])
temp_value = max(temp_value, temp_value_2) # use max score
if temp_value > limit:
out_imgpaths.append(_imgpaths[n])
similar_scores.append(temp_value)
# 查询es数据库，返回similar_infos
similar_infos = []
for id in range(len(out_imgpaths)):
id_img_str = out_imgpaths[id]
str1 = id_img_str[0]
str2 = id_img_str[1]
# 根据图像URL（视为唯一id），模糊匹配title，在解析title信息
print('==========', str2)
body = {"query": {"match": {str1: str2}}}
info_org = es_db.search(index='fwnews', body=body)
info_dst = info_org.get('hits').get('hits')[0].get('_source')
img_url = str2
id_title = info_dst.get('id')
all_imgs = info_dst.get('allimage')
title_url = info_dst.get('source_url')
web_source = info_dst.get('source')
title = info_dst.get('title')
img_similar_score = similar_scores[id]
similar_infos.append({
'img_url': img_url,
'title': title,
'title_url': title_url,
'web_source': web_source,
'img_similar_score': img_similar_score,
'id_title': id_title,
'all_imgs': all_imgs
})
if len(similar_scores) == 0:
return similar_infos, 0
else:
return similar_infos, max(similar_scores)
def load_img(orgimgroot):
hash_str_list = []
imgpath_list = []
orgimg_list = os.listdir(orgimgroot)
for n in orgimg_list:
imgpath = os.path.join(orgimgroot, n)
try:
img = cv2.imread(imgpath)
except:
print('read img error: ', imgpath)
img = None
img_hash = fun_Hash(img)
hash_str_list.append(img_hash)
imgpath_list.append(imgpath)
return hash_str_list, imgpath_list
def upimgs(upimgroot, orgimgroot):
_hash_strs = []
_imgpaths = []
upimg_list = os.listdir(upimgroot)
for n in upimg_list:
imgpath = os.path.join(upimgroot, n)
copy_path = os.path.join(orgimgroot, n)
shutil.copyfile(imgpath, copy_path)
os.remove(imgpath)
try:
img = cv2.imread(copy_path)
except:
print('read img error:', copy_path)
img = None
img_hash = fun_Hash(img)
_hash_strs.append(img_hash)
_imgpaths.append(copy_path)
print('adding: ', copy_path)
return _hash_strs, _imgpaths
