High quality Landsat 5 mosaic image production. After running this script, open https://code.earthengine.google.com/ and see tasks running procedure in the task window.

HQComposite.py

# Code for generating High-quality Landsat mosaics
# generated by Huan Li (huan.li@pku.edu.cn)
# Many thanks to the excellent GEE community for sample code sharing

# after run this script, open https://code.earthengine.google.com/ and see tasks running procedure in the task window
import ee
import time

ee.Initialize()

# global variables
NON_FILL = True
NON_SAT = True
CONFIDENCE = 1 # low 0 medium 1 high 2
SHADOW = True
CLIP_EDGE = False
EDGE_LENGTH = 7 # km
COMPOSITE = 1 # 0 mosaic top layer; 1 median mosaic; 2 mean mosaic
PARADIGM = 'pixel' # 'pixel' or 'scene'

# Function to clip image
def clip(im, distance):
    geo = ee.Geometry(im.geometry());
    small_geo = geo.buffer(ee.Number(distance).multiply(-1000));
    return im.clip(small_geo);

# Mask bad pixels
def maskBad(img):
    radsat_qa = img.select('radsat_qa');
    qa = img.select('pixel_qa');

    # fill
    fillPixel = radsat_qa.bitwiseAnd(1); # fill bit
    if NON_FILL:
        img = img.updateMask(fillPixel.Not())

    # saturated bands 1-7
    satPixel = radsat_qa.bitwiseAnd(1 << 1)\
        .Or(radsat_qa.bitwiseAnd(1 << 2))\
        .Or(radsat_qa.bitwiseAnd(1 << 3))\
        .Or(radsat_qa.bitwiseAnd(1 << 4))\
        .Or(radsat_qa.bitwiseAnd(1 << 5))\
        .Or(radsat_qa.bitwiseAnd(1 << 6))\
        .Or(radsat_qa.bitwiseAnd(1 << 7))\

    if NON_SAT:
        img = img.updateMask(satPixel.Not())

    # cloudy
    cloudPixel = qa.bitwiseAnd(1 << 5) # cloud bit
    if CONFIDENCE == 0: # low confidence
        cloudPixel = cloudPixel.And(qa.bitwiseAnd(1 << 6))
    elif CONFIDENCE == 1: # medium confidence
        cloudPixel = cloudPixel.And(qa.bitwiseAnd(1 << 7))
    else: # high confidence
        cloudPixel = cloudPixel.And(qa.bitwiseAnd(3 << 7))
    img = img.updateMask(cloudPixel.Not())

    # shadow
    shadowPixel = qa.bitwiseAnd(1 << 3) # shadow
    if SHADOW:
        img = img.updateMask(shadowPixel.Not())

    # edge
    if CLIP_EDGE:
        img = clip(img, EDGE_LENGTH);

    # Remove edge pixels that don 't occur in all bands  # 0 invalid, 1 valid
    edgeMsk = img.mask()\
        .reduce(ee.Reducer.min()) # minimum of all bands

    return img.updateMask(edgeMsk) # edge pixels

def ndvi(img):
    return img.addBands(img.normalizedDifference(['nir', 'red']));

def CompositeStrategy(imgCol):
    if COMPOSITE == 0:
        return imgCol.mosaic();
    elif COMPOSITE == 1:
        return imgCol.median();
    elif COMPOSITE == 2:
        return imgCol.mean();
    elif COMPOSITE == 3:
        imgColNDVI = imgCol.map(ndvi)
        return imgColNDVI.qualityMosaic('nd');
    else:
        return imgCol.qualityMosaic('temp');

def HQComposite(imgCol):
    maskedImgCol = imgCol.sort('CLOUD_COVER', False).map(maskBad);
    return CompositeStrategy(maskedImgCol)

def export_oneimage(img, folder, name, region, scale, crs):
    task = ee.batch.Export.image(img, name, {
        'driveFolder': folder,
        'driveFileNamePrefix': name,
        'region': region,
        'scale': scale,
        'crs': crs,
        'maxPixels': 1e11
    })
    task.start()
    while task.status()['state'] == 'RUNNING':
        print 'Running...'
        # Perhaps task.cancel() at some point.
        time.sleep(10)
    print 'Done.', task.status()

LT45_BANDS = ['B1', 'B2', 'B3', 'B4', 'B5', 'B7', 'B6', 'radsat_qa', 'pixel_qa'];
# LE7_BANDS = ['B1', 'B2', 'B3', 'B4', 'B5', 'B7', 'B6', 'radsat_qa', 'pixel_qa'];
# LC8_BANDS = ['B2', 'B3', 'B4', 'B5', 'B6', 'B7', 'B10', 'radsat_qa', 'pixel_qa'];
STD_NAMES = ['blue', 'green', 'red', 'nir', 'swir1', 'swir2', 'temp', 'radsat_qa', 'pixel_qa'];

# Poyang lake: 115.4745,28.2953,116.955,29.8416
Poyang = ee.Geometry.Rectangle(115.4745,28.2953,116.955,29.8416);
ee.Geometry.Polygon([[[34, -2],[34, -3],[35, -3],[35, -2]]])
imgCol = ee.ImageCollection('LANDSAT/LT05/C01/T1_SR').select(LT45_BANDS, STD_NAMES)\
    .filterDate('2010-01-01', '2010-12-01')\
    .filterBounds(Poyang)                      # Set temporal extent and spatial extent here
# print(imgCol.first().getInfo())

# mosaicImg = imgCol.mosaic().clip(Poyang).toFloat().getDownloadUrl({
#         'scale': 1000,
#         'crs': 'EPSG:4326',
#         'region': '[[115.4745, 29.8416], [116.955, 29.8416], [116.955, 28.2953], [115.4745, 28.2953]]'
# # })
# Producing high quality mosaic image
if PARADIGM == 'pixel':
    mosaicImg = HQComposite(imgCol);
else:
    mosaicImg = CompositeStrategy(imgCol.sort('CLOUD_COVER', False));
# print(mosaicImg)

# export parameters
outputFolder = 'HQExport'
outputFileName = 'L5_exportImg'
scale = 1000            # modify image resolution
crs = 'EPSG:4326'
# set export region
region = str([[115.4745, 29.8416], [116.955, 29.8416], [116.955, 28.2953], [115.4745, 28.2953]])

# export
export_oneimage(mosaicImg, outputFolder, outputFileName, region, scale, crs)