def search_signatures(indexname, level_filenames, feature_filenames):
import correlate
import smooth_avg
level_signatures = []
for level_filename, feature_filename in zip(level_filenames,
feature_filenames):
level_signature = signature_read.read_signature_short(level_filename)
level_signatures.append(level_signature)
data = pickle.loads(open(indexname, "r").read())
search_results = []
for level_filename, feature_filename in zip(level_filenames,
feature_filenames):
level_signature = signature_read.read_signature_short(level_filename)
feature_signature = signature_read.read_signature_byte(
feature_filename)
signature = signature_make.signature_make(level_signature,
feature_signature)
search_result = signature_make.signature_search_intersect(
data, signature)
search_result = [
(i, compare,
correlate.pearson(smooth_avg.avg_smooth(level_signature),
smooth_avg.avg_smooth(level_signatures[i])))
for i, compare in search_result
]
# search_result = [(i, compare) for i, compare in search_result]
search_results.append(search_result)
return search_results
def signature_make(level_signature, feature_signature):
result = []
level_range = 2**16
feature_range = 2**8
import smooth_avg
level_signature = smooth_avg.avg_smooth(level_signature)
for (last_level, last_feature), (level, feature) in adjacent(zip(level_signature, feature_signature)):
if abs(level - last_level) > level_range / 8:
feature_value = last_feature * feature_range + feature
result.append(feature_value)
return result
def signature_make(level_signature, feature_signature):
result = []
level_range = 2**16
feature_range = 2**8
import smooth_avg
level_signature = smooth_avg.avg_smooth(level_signature)
for (last_level, last_feature), (level, feature) in adjacent(
zip(level_signature, feature_signature)):
if abs(level - last_level) > level_range / 8:
feature_value = last_feature * feature_range + feature
result.append(feature_value)
return result
def window_normalize(values, window = 200):
import smooth_avg
values = smooth_avg.avg_smooth(values)
forward_avg, backward_avg = [], []
queue = deque(maxlen = window / 2)
for value in values:
forward_avg.append((sum(queue), len(queue)))
queue.append(value)
queue = deque(maxlen = window / 2)
for value in reversed(values):
backward_avg.append((sum(queue), len(queue)))
queue.append(value)
avg = [float(fs + value + bs) / float(fl + 1 + bl) for value, (fs, fl), (bs, bl) in zip(values, forward_avg, backward_avg)]
return [value / average if average else 0 for value, average in zip(values, avg)]
def search_signatures(indexname, level_filenames, feature_filenames):
import correlate
import smooth_avg
level_signatures = []
for level_filename, feature_filename in zip(level_filenames, feature_filenames):
level_signature = signature_read.read_signature_short(level_filename)
level_signatures.append(level_signature)
data = pickle.loads(open(indexname, "r").read())
search_results = []
for level_filename, feature_filename in zip(level_filenames, feature_filenames):
level_signature = signature_read.read_signature_short(level_filename)
feature_signature = signature_read.read_signature_byte(feature_filename)
signature = signature_make.signature_make(level_signature, feature_signature)
search_result = signature_make.signature_search_intersect(data, signature)
search_result = [(i, compare, correlate.pearson(smooth_avg.avg_smooth(level_signature), smooth_avg.avg_smooth(level_signatures[i]))) for i, compare in search_result]
# search_result = [(i, compare) for i, compare in search_result]
search_results.append(search_result)
return search_results
def slide_normalize(values):
import smooth_avg
values = smooth_avg.avg_smooth(values)
values = [value + 2**15 for value in values]
forward_avg = []
backward_avg = []
alpha = 0.5
average = 0.0
for value in values:
average = average * alpha + value * (1 - alpha)
forward_avg.append(average)
average = 0.0
for value in reversed(values):
average = average * alpha + value * (1 - alpha)
backward_avg.append(average)
backward_avg.reverse()
bi_avg = [(forward + backward) / 2 for forward, backward in zip(forward_avg, backward_avg)]
result = []
for value, average in zip(values, forward_avg):
if value and average:
result_value = value / average
else:
result_value = 0
result.append(result_value)
return result[int(len(result) * 0.1):-int(len(result) * 0.1)]
