Visualization of BBOB functions

Recovering of the landscapes of selected functions from the black-box optimization benchmark (BBOB).
We used f17-f24 for demonstrations here, which exhibit high level of multimodality and are more complicated (we also show f1, which is a unimodal one for illustration of easier ones).
The task tested is to recover the landscape of each of these 2d functions. We first draw grid samples from the search space (40*40), and then create fitness landscapes based on Euclidean distance with a neighborhood distance threshold of 0.5. 
We then visualize the landscape using our proposed method, and compares it to the groundtruth function surface.