Automatic detection of surface defects on fruit by using a vision system

Abstract

The OECD standards for the fresh market make it necessary to estimate the quality of fruit by grading by size, colour and surface defects. While size and colour grading are now automated, sorting of damaged fruit is still done manually. A system for automatic detection of surface defects on Golden Delicious apples has been developed and tested on line, on a conveyor system used for automatic colour grading. While the fruit is rolling, a solid-state camera takes four pictures, and so is able to view most of the surface. A hardware system has been developed to reduce the amount of data to be processed. This is based on a model which assumes the apple to be spherical with variable diameter, according to the fruit size. The system can analyse more than five fruit per second per grading line. The tests on line showed that 69% of the fruit were correctly graded, but 26% were classified immediately above or below the right grade. The rough model employed allowed only marketable fruit (grades 1 and 2) to be distinguished from the unmarketable fruit (grades 3 and waste), and the grade 3 to be distinguished from the waste. Consequently, this system can be used only during the picking period, in order to sort the fruit unmarketable as fresh fruit. However, it can also be used for the pre-size grading operations before storage, since it gives the weight with a standard deviation of ± 5 g ( r = 0·985).