Zebrafish have emerged as a powerful model organism in cardiovascular disease research. Accurately identifying zebrafish blood vessels and evaluating blood flow velocity without injury has a wide range of biological applications. This paper presents the design and development of a non-invasive microvision system for vascular recognition and blood flow monitoring of zebrafish larvae. For the first time, a visual algorithm based on color thresholding and discrete Fourier transform filtering is proposed to determine the position of zebrafish dorsal cardinal vein vessels. Next, the blood flow velocity is determined based on the change rate of pixel values near the centroid point of the blood vessel recognition results. Then, an independent software system is developed based on the producer-consumer underlying framework. A user-friendly interface is specifically designed for biomedical workers, and a complete prototype system is built in combination with hardware devices. In addition, relevant experiments were conducted, and the results indicated that the system can effectively recognize the position of vessels and monitor blood flow velocity in zebrafish larvae under different anesthesia concentrations and developmental days. The heart rate information obtained based on blood flow velocity is consistent with the heart beating frequency. Moreover, the system has also been successfully applied to blood flow velocity monitoring under fluorescence conditions. In future work, this system will be applied in drug screening research for cardiovascular-related diseases of zebrafish larvae.