tarohi24 / notebooks

What I learned
1 stars 0 forks source link

Similarity between images #28

Open tarohi24 opened 4 years ago

tarohi24 commented 4 years ago

References:

tarohi24 commented 4 years ago

There are two categories of similarity search:

due to the application to my research, I focus on CBIR

tarohi24 commented 4 years ago

Average hash

the basic algorithm

  1. gray scaling
  2. shrink images
  3. calculate the average of pixels
  4. binary hashing. if a pixel is above the average, it's 1. otherwise it's zero.

just like an ascii art.

1111111111001111
1111111110000111
1111111100100111
1111111100110011
1111111101110011
1111111100110011
1111111100000011
1101011111100011
1111111110100011
1111111110000111
1111111110000011
1111111110000001
1111111100000000
1111111000000000
1111110000000000
1111100000000000