s1091536 作業6
1. 題目說明：
Run-Length
Based Image Compression 練習
附件中為三張利用將晶片高度以色彩視覺化後的圖片。 請設計一個基於
Run-Length 的壓縮法方，對圖檔作無失真壓縮後儲存成新檔案。 部落格上應敘述你的壓縮方法，提供壓縮檔之格式，並計算三張圖的平均壓縮率 (compression ratio)。
2.開發環境：
OS: Windows 11
Editor: PyCharm
Language: python 3.8
Package: OpenCV 4.7.0
3.操作：
“python .\hw6_1091536.py”啟動程式。
將壓縮後的資料存成.txt，並產生解壓縮後的檔案.bmp
程式依序輸出：
壓縮完成、壓縮完成、解壓縮成功(是否與原始檔案相同)
壓縮前檔案大小、壓縮後檔案大小、壓縮率、資料減少量比率
剩餘資料量率
=>
=>
val, 次數, val, 次數, val, 次數, val, 次數, val, 次數, val, 次數, ...
結果輸出：
-------------------------
img1 -------------------------
img1
is encoding...
Encoding
done!
Decoding
done!
Decode
success!
img1 size:
Before: 14321.537109375 KB
After: 5800.8671875 KB
Compression
ratio: 2.469
Relative
data redundancy: 59.5 %
###
Remain 40.5 % of data ###
-------------------------
img2 -------------------------
img2
is encoding...
Encoding
done!
Decoding
done!
Decode
success!
img2 size:
Before: 14321.537109375 KB
After: 9990.9765625 KB
Compression
ratio: 1.433
Relative
data redundancy: 30.2 %
###
Remain 69.8 % of data ###
-------------------------
img3 -------------------------
img3
is encoding...
Encoding
done!
Decoding
done!
Decode
success!
img3 size:
Before: 14321.537109375 KB
After: 5322.6923828125 KB
Compression
ratio: 2.691
Relative
data redundancy: 62.8 %
###
Remain 37.2 % of data ###
Average Compression ratio: 2.198
4.方法及實作：
(a)概念&程式：
讀入影像
img1 =
cv2.imread('img1.bmp')
壓縮
def RLencoding(img, path): # encode
(rows, cols, channels) = img.shape
# split 3 channels to 3 arrays
B, G, R = cv2.split(img)
print(path, 'is encoding...')
path += '.txt'
f = open(path, 'w')
# flatten each channel to 1 dimension
B = B.flatten()
f.write(str(rows) + "," + str(cols) + "," + "\n")
# encoding for B
val = B[0]
quantity = 0
for i in range(len(B)):
if B[i] == val:
quantity += 1
else:
f.write(str(val) + "," + str(quantity) + ",")
val = B[i]
quantity = 1
f.write(str(val) + "," + str(quantity) + ",")
f.write("\n")
# encoding for G & R
...
f.close()
print('Encoding done!')
解壓縮
def RLdecoding(path): # decode
rows, cols = 0, 0
B = np.array([])
...
with open(path) as f:
count = 0
for line in f.readlines():
t = line.split(',')
t.pop()
t = list(map(int, t))
if count == 0:
rows = t[0]
cols = t[1]
elif count == 1:
B = t
elif count == 2:
G = t
else:
R = t
count += 1
B = np.array(B)
...
# reshape to 2 dimension
B = B.reshape(int(len(B) / 2), 2)
...
# transform to image format
b = input_to_img(rows, cols, B)
...
output = np.array(cv2.merge([b, g, r]), dtype='uint8')
print('Decoding done!')
return output
留言
張貼留言