Abstract: Intelligent perception of environmental parameters and fish status in marine breeding was the primary condition for the transformation and upgrading of the marine breeding industry. With the development of marine breeding from near to far and from shallow to deep, many scholars had begun the study of optical technology in marine breeding. However, there were fewer applications for optical technology in marine breeding. Therefore, it was necessary to comprehensively review this work to promote the development of optical technology in marine breeding. Optical technology could collect perception information for the wisdom fisheries of marine pastures. It was an important underwater sensing technology, but it was still limited in the application. This thesis used the influence mechanism of the light environment on fish as the starting point and summarized the positive role of halo regulation in different growth periods such as fish -based gonad development, incubation phase, seedling stage, and growth stage. During the different growth period of fish, appropriately regulate light environmental elements such as light intensity, light cycle, light source frequency, and light color, which has a positive effect on the breeding and growth of fish seedlings. By regulating the elements of the light environment, the growth process of marine breeding fish is controllable and increases the efficiency. This paper focused on the engineering application effect of the detection algorithm and combing the application of computer vision technology in fish growth and behavior recognition. Based on underwater computer vision technology, it can master the size, quality and biomass of farming, and monitor the speed and trajectory of fish healthy swimming, the degree of fish hunger and feeding behavior, the change of fish body colour and fish cope with the breeding environment Abnormal stress behavior, etc. It is very important for the marine breeding expert system. In real -time monitoring of fish behavior status by computer vision, timely finding that fish are in the state of hunger or stress reaction behavior and predict early warning, which can reduce output loss and effectively maintain the safety of breeding assets. This thesis summarized the spectral method from multiple angles such as the degree of light, fluorescence, and high spectrum. The real -time online water quality parameter monitoring of high reliability, high stability, and high accuracy is realized through spectral methods and online fish epidemic detection of actual breeding environment. It is an important technical method to ensure green breeding and far -reaching sea farming. Finally, this thesis discussed and looked forward to technical challenges and development trends of optical technology in turbid water imaging, fish behavior status recognition, and online testing of fish epidemic. This thesis explored hard software methods to improve the quality of optical imaging of muddy water. At the same time, this thesis explored the innovation direction of fish behavior recognition technology of the fusion of sound and light. In addition, this thesis explored the innovative direction of hyperspectral detection in fish epidemic detection and the development trend of online detection technology for seawater micro components in underwater aquaculture. The systemic improvement of optical technology in marine breeding can provide high -level texture, intuitive three-dimensional, multi-dimensional, high-precision and powerful intelligence technical support for the smart fisheries of marine pastures.