Case Study: Fruit Classification

A supplier of tomatoes needs to be able to classify batches of fruit based on the quality of the product. 

An overview

The product, batches of tomatoes, are presented on a slow moving conveyor. The tomatoes are over a layer deep. The producer will assign the batch depending on its quality. It is not necessary to inspect each and every tomato but instead to sample the the top layer of fruit and assign the batch accordingly.

The batch is to be classified based upon the ripeness of the fruit and the size of the fruit. The type of fruit, and therefore the shape of fruit (plum, cherry and so on) is known.

Two high resolution GigE colour cameras are mounted on an xy stage above the tray. The images from each of the cameras is stitched together. Each acquisition covers an area of 600mm x 450mm. The XY stage constantly traverses the slow moving conveyor of tomatoes, acquires images and generates information allowing the producer to trend the colour & size of the fruit and the portion which is unripened.

The area is illuminated by a diffuse array of leds. The camera looks through a hole in the array, The illumination moves on the stage with the leds.

The image is processed in order to more clearly define the boundary between tomatoes. The segmentation of the fruit is performed on the processed imaged and the locations of the best candidates projected back onto the original colour image.

Once acquired the images are passed to an industrial PC for processing. The two images are stitched together. The image is optimised for segmentation by applying a number of filter and arithmetic operations.

Finally a segmentation routine identifies candidate tomatoes and a sorting algorithm removes all but strong matches.

Each located fruit is then inspected for colour, consistency of colour and size.

Take the Next Step

Contact us today and let's discuss how we can turn your vision into reality. Together, we'll design, develop, and deploy your vision system better.