A Survey of Potatoes Image Segmentation Based on Machine Vision

Bi-Level Image Segmentation

In order to obtain a simple and fast system, monochrome (grayscale) images are the best choice (Russ, 2016). Using grayscale images reduces the complexity and speeds up the system. There are different works of using this kind of system. The general algorithm of this approach is as follows:

• 1.

  Image acquisition

• 2.

  Convert image into the gray color space

• 3.

  Image thresholding

• 4.

  Performing the morphological operations

• 5.

  Two level segmented image including potato tuber and background

In brief, after reading the image, it should be first converted to the gray color space. Figure 1 shows the gray mode for a simple image.

Figure 1.

(A) Input image (RGB color space) and (B) gray conversion of the (A)

In Figure 2, the gray image should be threshold by a threshold technique to achieve a bilevel image including the potato and background. For instance, consider the previous image; by applying a threshold method (for example Otsu method is analyzed (Xu, Xu, Jin, & Song, 2011)) and the morphological operations, the output image including a highlighted potato tuber separated from its background is achieved.

Figure 2.

Grayscale image (A) and its threshold result after image threshold and morphological operations (B)