A fish image processing system and method

Abstract

The invention relates to a fish image processing method, which comprises the following steps of: obtaining a first fish image which comprises fish information; according to the rain and fog removal mathematical model,establishing a rain and fog removal network, and performing rain and fog removal on the obtained first fish image by using the rain and fog removal network to obtain a second fish image; Wherein B refers to a background image without any interference. Wherein the image represents a rain image layer, O represents an actually recorded image, R represents whether rainwater exists ineach pixel of the actually recorded image, A represents atmospheric background light, and (1-alfa) is an attenuation ratio in the process of propagating reflected light on the surface of an object toa camera; And processing the second fish image after rain and fog removal by using a fish detection algorithm, and marking the position of the fish in the second fish image. The method has the beneficial effects that the rain removal operation and the fog removal operation are completed in the same stage through a deep learning method, the position of the fish is effectively positioned through a fish detection algorithm, the category is marked, and the fish recognition efficiency is improved.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a fish image processing system and method. Background technique [0002] Illegal and unsustainable fishing is common in all regions of the world and has had a significant impact on marine ecosystems around the world. On the one hand, the number of many fish stocks represented by bigeye tuna has fallen below the warning line; on the other hand, nearly 50% of the world's population uses seafood as their main source of protein, and the economies of many central and western Pacific island countries such as Nauru rely heavily on tuna, etc. Marine life, enhanced regulation to achieve sustainable development is imminent. [0003] Since the end of the last century, many countries and international organizations have installed surveillance cameras on the decks of fishing boats, hoping to prevent fishermen from catching inappropriate fish at inappropriate seasons. The...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problems of water mist interference and difficulty in fish identification in the prior art, the present invention provides an image processing system and method, which has the characteristics of being able to eliminate rain and mist interference and effectively identify fish, etc.

Images