Algorithm 1 KNN-based pixel accumulation algorithm.

Pixel Accumulation Algorithm Based on KNN Input: Single photon matrix X, Label matrix Y, Ground truth matrix Z 1:   for ${\mathit{z}}_{i,j}$ = ${\left[z_{i,j}^1,z_{i,j}^2,z_{i,j}^3,\dots ,z_{i,j}^h\right]}^T$ in Z do 2:      ${\mathit{f}}_{i,j} \underleftarrow{featureextraction} {\mathit{z}}_{i,j}$ 3:   end for 4:   P = $\left\{(\right.$fi,j, yi,j)} 5:   for ${\mathit{x}}_{i,j}$ = ${\left[x_{i,j}^1,x_{i,j}^2,x_{i,j}^3,\dots ,x_{i,j}^h\right]}^T$ in X do 6:      ${\mathit{f}}_{i,j}^x$$\underleftarrow{featureextraction}$${\mathit{x}}_{i,j}$ 7:   end for 8:   B = $\left\{{\mathit{f}}_{i,j}^x\right\}$ 9:   for ${\mathit{f}}_{i,j}^x$ in B do 10:      NK$\left({\mathit{f}}_{i,j}^x\right)$← K nearest points in P 11:      ${\mathit{y}}_{i,j}^x$ = $\underset{c_{i,j}}{argmax}{\displaystyle \sum _{p_{i,j}\in {\mathit{N}}_K\left({\mathit{f}}_{i,j}^x\right)}}\phi \left(y_{i,j},c_{i,j}\right)$ 12:   end for 13:   for ${\mathit{x}}_{i,j}$ = ${\left[x_{i,j}^1,x_{i,j}^2,x_{i,j}^3,\dots ,x_{i,j}^h\right]}^T$ in X do 14:      if ${\mathit{y}}_{i,j}^{x^{\prime }}$∈${\mathit{N}}_8$(${\mathit{x}}_{i,j}$) = ${\mathit{y}}_{i,j}^x$ then 15:         ${\mathit{q}}_{i,j}$ = ${\mathit{x}}_{i,j}$ + ${\mathit{x}}_{i,j}^{\prime }$ 16:      end if 17:   end for Output: Data matrix Q