Automatic image segmentation using sobel operator and k-means clustering: A case study in volume measurement system for food products

Abstract

Abstract-Image segmentation plays an important role in automatic visual inspection of food product using computer vision system. However, segmentation of food product image is not easily performed if the image has low contrast with its background or the background in acquired image is not homogeneous. This paper proposes a method for automatic food product image segmentation using Sobel operator and k-means clustering. Sobel operator was used to determine region of interest (ROI). k-means clustering was then used to separate object and background in ROI. The area outside ROI was considered as background. The proposed method has been validated using 100 images of food product from ten different types. The experimental results show that the proposed method achieves good segmentation result.

Computer vision system for egg volume prediction using backpropagation neural network

“…The darker shell color would be caused by some pixels near the boundary of egg recognized as background in segmentation process. To overcome this drawback, a local segmentation, as proposed by [13], can be applied to replace automatic thresholding. For comparison, the volume of each sample was also measured using volume prediction method proposed by [1] and obtained mean square error of 26.1226, correlation coefficient of 0.7281, and mean ARE of 3.5016%.…”

Computer Vision Estimation of Physical Parameters and Its Application to Power Requirements of Natural and Artificial Swimmers

“…There has been considerable research done in the realm of fish classification and estimation of physical dimensions using computer vision. Traditional image segmentation and volume estimation methods are presented by Siswantoro et al; these authors used k-means clustering and the Sobel operator [26]. Balaban et al measured Alaska pollock (Theragra chalcogramma) by taking side and top view images and then estimating the body contour as a b-spline.…”

“…The Python OpenCV package was used to process the images and determine body geometry contours [45]. The image segmentation algorithm used was an adapted version of the automatic image segmentation algorithm given in Siswantoro et al [26], and an overview of the adapted algorithm is shown in Figure 3. When an image is read into the program, it is read as a matrix with dimensions (H, W, C), where H and W are the height and width of the image and C represents a vector of three color channels (blue, green, red) as in Figure 4a.…”

Robotic Aubergine Harvesting Using Dual-Arm Manipulation

“…The techniques commonly used in image segmentation are thresholding-based, gradient-based, region-based, edge-based, and classification-based [25]. Within the classification-based techniques, machine learning and deep learning algorithms play a relevant role by establishing relationships among multiple features to improve system efficiency.…”