Adaptive Bacteria Colony Picking in Unstructured Environments Using Intensity Histogram and Unascertained LS-SVM Classifier

Algorithm 1

Entropy based mean shift filtering algorithm.

Let 𝑥_𝑖𝑖=1,…,𝑛 be the input image. Let 𝑂_𝑖𝑖=1,…,𝑛 be the filtered image. Pixel 𝑝∈𝑥_𝑖,𝑝=𝑅,𝐺,𝐵,𝑥,𝑦 ∈  R ⁵.

Let ent.0 be the entropy initial value, ent.1 be the next iteration of ent.0, and ent.2 be the next value of ent.1.

Let erras be the absolute value of the difference between the first two iterations. Let edset be the

thresholding as iterations stopping criteria. Our algorithm comprises the steps listed below:

 (1) Initialize ℎ=1, 𝑦_ℎ,1=𝑝_ℎ, 𝑒𝑛𝑡.2=1, 𝑒𝑟𝑟𝑎𝑠=1, 𝑒𝑑𝑠𝑒𝑡=0.

 (2) While 𝑒𝑟𝑟𝑎𝑠>𝑒𝑑𝑠𝑒𝑡, then

   (2.1) Filtering image based on mean shift. Store the image in 𝑂^[𝐾].

    (2.2) Calculating entropy from the 𝑂^[𝐾] and store the image in 𝑒𝑛𝑡.1.

    (2.3) Entropy is used to calculate the absolute difference which is obtained in the previous step; 𝑒𝑟𝑟𝑎𝑠=|𝑒𝑛𝑡.1−𝑒𝑛𝑡.2|.

    (2.4) Update the parameters; 𝑒𝑛𝑡.2=𝑒𝑛𝑡.1 and 𝑂^[𝑘+1]=𝑂^[𝑘].

    (2.5) Calculate the mean shift which is carried out until entropy convergence.

 (3) Store 𝑍_𝑖 which is calculated 𝑍_𝑖=(𝑥^𝑠_𝑖,𝑦^𝑟_𝑖,𝑐), here  𝑥^𝑠_𝑖   is the spatial information and 𝑦^𝑟_𝑖,𝑐  is the color range information.