Abstract: In view of the low efficiency of the traditional sorting method of blueberry fruit，the digital recognition and calculation of the projection area of blueberry fruit were proposed by using machine vision technology. The image of blueberry fruit was divided by the maximum inter-class variance method，the connected domain area was removed by morphological processing method，the blueberry area was approximately rounded by the least-square method，and the area and perimeter of blueberry fruit were calculated after calibration. The results showed that in the case of manual segmentation results as the standard for calculating relative accuracy，the use of the blueberry area as the grading standard was more accurate than the use of perimeter as the grading standard，and the effect was better. After the image processing，the average accuracy of the calculated area reached 98. 93%，the effect was stable，with high accuracy. It provides an important basis for the automatic grading of blueberry fruit.

Key words: machine vision, blueberry fruit, least-squares fitting, image processing