Guava Bruise Area Calculation Using Color And Grayscale Image Segmentation

A bruise area calculation method on guava fruit surface was presented that relied on image segmentation by identifying boundaries or edges appearing in the guava image. Boundary detection was used to calculate the entire area of the guava and this was then compared with the bruise area. Image data presented as color combination of R, G and B was difficult to apply and performed poorly, while simple foreground and edge segmentation methods using grayscale images provided good results. The bruise region had maximum image intensity and the thresholding image segmentation method was used to specify this area. Results were not acceptable because of sensitivity to light and shadow. A color image segmentation method based on thresholding was used to overcome this issue. Two color spaces, HSV and L*a*b, clearly distinguished between the color of the guava and the color of the bruise. A specific portion of the covered color area could be specified in both color spaces and both produced good results. After specifying the desired region, the calculated area was compared between the grayscale and color models. Color image segmentation by the color thresholder produced good results that matched our data.