Traditionally food inspection has been done through human inspection a process which was tedious, expensive, destructive, and most a time wasn’t reliable given the uncountable human errors. Currently, the use of technology for food inspection has been introduced globally and has proven effective. The increased consumer awareness and quality demand in the market have created the need for quality sorting and delivery of top quality products in the market for it is a necessity not just for consumer taste but also a need for human health. Computer vision process provides an alternative for a computerized, non-destructive, cost-effective technique to accomplish the task. The process currently is applied to assessing the quality of fruits, vegetables, and eggs, and other food products. The process includes capturing, processing, and analyzing images, facilitating the objective and non-destructive assessment of the visual quality characteristic of food products.
The basics of computer vision quality check on food products.
A digital image is produced by several image sensors which include many light-sensitive digital cameras, range sensors, radar, tomographic sensors, ultrasonic cameras amongst others. Depending on the quality of the sensors used the image produced can be 2D or 3Dimage or image sequences.
Preprocessing.
This is a process used to determine whether the images contain specific data required for the process. The process includes Resampling, to assume that the image coordinates system is correct. Noise reduction to ensure that the sensor noise does not cause false information, contrast enhancement to ensure all details are detected, and scale-space representation to ensure that image structures are at a locally appropriate scale.
Feature extraction.
Specific features are extracted from the images which will help determine the quality of a product. This feature may include line edge and ridges or localized interest points such as bolds and points while more complex features may be related to motion, shape, or texture.
#machine-learning #computer-science #tensorflow #data-science
