To reduce the ellipticity of the steel pipes processed by JCO forming,a precise bending technology was presented based on plastic bending engineering theory and machine vision technology.In the first bending step,the prescribing two-bend was used to obtain the springback law,then using the springback law and considering the aimed bend angle, the exact punch displacement to determine the forming angle in each bending step was predicted.In the succedent bending steps, the bending process was carried out with the exact punch displacement predicted above.The angle error in each bending step was calculated by comparing the actual forming angle with the aimed bend angle, and each precise bending process was realized by compensating for the last angle error in the next bending step.Furthermore, the image processing and line detecting algorithms suitable for the image of the steel pipe end plane were proposed by a machine vision system.A calibration method was presented to calculate the exterior parameters of the CCD camera, which employed a equilateral triangle as the calibration board and only needs to capture one image.A mathematical model was derived to convert the angle in the image into the actual forming angle.Experimental results show that the measuring error of the forming angle is less than 0.2° and the ellipticity of the formed pipe is less than 1.5%.Employing the precise bending technology, the high-quality pipes can be manufactured without depending on worker’s operating experiences.