In order to solve the problems of long time-consuming, low precision, low degree of automation, and inadaptability to multiple variety counting in the current crop 1000-grain weight measurement, a 1000-grain weight measuring instrument with the function of weighing and counting was designed based on machine vision. The device is mainly composed of a frame, a camera, a loading plate, a weighing sensor, a controller and other components. At the same time, seed images are collected by the camera and uploaded to the upper computer for image analysis and processing to obtain the number of seeds. Finally, the mass of thousands of seeds is obtained by automatic calculation and the results are displayed on the software interface so that the operator can obtain the measurement results in time. In order to realize the image segmentation of adhesive seed to get accurate seed number, the seed images were processed and analysis with extracting the seed region by removing the border, preprocessing the seed region by using median filtering, maximum inter-class variance method and morphological operation, and finally segmenting the image by using the watershed algorithm based on distance transformation, and obtaining the seed number by counting the number of connected regions. With rapeseed, maize, rice and soybean as the experimental subjects, 100 groups of counting, weighing and 1000-grain weight determination experiments were carried out. The results showed that the number error was no more than 1 grain. Weighing accuracy was 99.86%, 99.92%, 99.86%, 99.96%, respectively. The accuracy of 1000-grain quality were 99.93%, 99.63%, 99.61%, 99.97%, respectively. The time of 1000 grain mass determination was less than 2 s.