The Research Of Automatic Fry Separation System By Computer Vision

In the fry breeding process, large fry will attack small fry in a pond, small fry will be injured or even death, resulting in economic losses, so the fry separation is necessary. For the low efficiency of traditional manual separation methods and lacking of scientific guidance, this paper proposed a fry automatic separation system by computer vision which can calculate fry length and get classified information according to the fry image. Error of length measurement accuracy is mainly due to the fish body bent, for this situation proposed a method to calculate the curved fry length by the back image with a high degree of accuracy. Through comparing with length results calculate by side measurement, the method in the paper has a high accuracy. The system can separate fry successfully and solve practical problems.The automatic fry separation system by computer vision combines with automation technology and computer vision technology, which can solve many problems caused by traditional methods. The main contents are as follows:(1) Establish a hardware automation fry separation platform: through a combination of computer vision technology and industrial technology, designed a platform which can achieve fry acquisition, fry image acquisition, fry length measurement by image processing, fry classification and separation guide.(2) Establish a software platform: Using digital image processing techniques to measure curved fry length and classify the fry. The fry length information will be measured by a series of digital image processes for fry image. As well as, fry can be separated by comparing the threshold information in database.(3) Establish a robust access control: Setting permissions for different users, so that have different rights that can operate system with different functions. That can operate system safely.