The sensing technology of vehicle is the primary problem for advanced driver assistant system(ADAS) and autonomous driving functions. The fusion of millimeter wave radar and camera is an important trend to enhance the environmental perception performance. Converting the multiple target parameters (such as distance, velocity, angle, etc.) detected by radar sensor into the corresponding image coordinates is the key of fusion algorithm processing, as well as a difficulty. This paper presents a method in order to detect and classify the on-road obstacles, like cars or not cars (other obstacles)，faster and more accurately. Adding a dimension corresponding relation between the radar cross section and pixel area in image, the multiple target parameters(R: range, θ:angle, v: velocity, RCS: Radar Cross Section) measured by the millimeter wave radar is converted to the corresponding image information ((x, y):the space coordinate of target, v: velocity in the image, A: pixel area) by the new rotation matrix Rt. The targets detection process can be divided into three stage, the first consist in reading radar signals and capturing the camera data, the second stage is the data frame synchronization and fusion, and the third stage is targets detection and parameters display. The experimental result demonstrates that this fusion arithmetic can enhance multiple targets detection performance, which provides the basis of target tracking process.