Which one can be expressed better using Huffman coding from those images

the red line is not part of the image. It has been used to indicate the border of the image

I'd say they are equally expressed since the frequency of the "black" and "white" pixels are equivalent. Huffman coding benefits from a distribution in the frequency of values within the image. The most frequent pixels are encoded using fewer bits and the least frequent pixels are encoded using more bits. In this case, you can use "0" and "1" to encode the pixels in both images. If there were more compression techniques applied such as run-length encoding then the right image would benefit more.

Black Pixels → 50% Frequency → Binary "0"

White Pixels → 50% Frequency → Binary "1"

Also in the case where there are only two-pixel intensities present Huffman encoding does not benefit from frequency variations.