Traditional methods measuring the gun’s jump angle have the disadvantages of low measurement efficiency, large error, weak practicability, and narrow measurement range. Aiming at these problems, a gun’s jump angle measurement method based on the improved least-squares algorithm is proposed. The method uses the improved circle fitting algorithm to detect the center position of the muzzle image, and obtains the coordinates of the gun shooting target point under the calibration of the binocular cameras. The target is collimated with the center position of the muzzle image and then shot. In data processing, direction of the gun initial velocity vector is calculated and used to obtain the actual measurement result of the gun’s jump angle, combined with the center position of the muzzle image and the aiming direction of the shooting target. The experimental results show that the improved algorithm is with fewer parameters, simple objective function, and low computational complexity, and it shortens the calculation time and improves the accuracy of the muzzle image center positioning. The measurement results show that the accuracy of the measured jump angle is within 0.5', which verifies the accuracy of the jump angle measurement system.