Abstract: This paper introduces a non-contact sensor for measuring object displacement; and elaborates on the key steps of image transformation, A/D conversion, post-processing data, and fieldbus technology. Verification shows that this solution is stable, reliable, accurate, and cost-effective. Keywords: Image transformation; A/D conversion; Image processing; CAN bus In some automatic control applications, it is often necessary to measure the width of gaps and crevices in real time for closed-loop automatic control. However, general image processing solutions suffer from drawbacks such as complexity, high cost, and large data processing volume. Based on the specific needs of industrial sites, a stable, reliable, accurate, and low-cost solution has been found. System Structure and Working Principle Figure 1 shows the principle block diagram of the test system. Figure 1 a. Image Acquisition Currently, various imaging systems generally use CCD image sensor technology, which has advantages such as low readout noise, large dynamic range, and high response sensitivity. However, image sensors based on CCD technology have disadvantages such as large size and high power consumption, making it difficult to achieve single-chip integration. With the development of CMOS process technology, the high integration of CMOS image sensors has reduced system complexity and manufacturing costs, making the readout and processing of acquired image information simple and fast. It also offers advantages such as small size, light weight, low power consumption, and low cost. Miniature cameras developed using CMOS image sensors can output PAL black-and-white video signals, serving as input components for computer acquisition systems. The most prominent advantage of miniature cameras is their exceptional sensitivity to infrared light, invisible to the human eye; their spectral sensitivity in the near-infrared band is 5-6 times higher than their sensitivity to visible light. In applications requiring continuous monitoring, it is necessary to equip the camera with an appropriate infrared illumination source, thus giving the system excellent environmental adaptability. A complete image acquisition device (see Figure 2) generally consists of a camera (1) and an infrared light source (2). The camera (1) is installed inside the housing (4) of the bracket (3). The front cover (5) of the housing (4) has a hole (6) that fits with the lens (7) of the camera (1). Several infrared light sources (2) are installed around the hole (6) on the end cover (5). The camera (1) has an output port for connecting to a subsequent computer image processing device. The infrared light source (2) is installed inside the hole (8) of the end cover (5). The distance H between the lens (7) and the end face is 6-10 mm. The mounting bracket (3) provides a convenient and accurate on-site assembly structure for the probe. Generally speaking, the gap environment in the industrial field is very poor, and it is difficult to directly measure the state of the gap. For gap areas with a measurement range of only a few millimeters, the resulting image will inevitably be blurry. In this case, the method used is to attach a ruler to the side of the indicator rod next to the gap, and fix the acquisition device in front of the ruler to obtain the image signal reflecting the state of the gap. Figure 2 b. Non-contact displacement detection In this scheme, we use a miniature black and white CMOS camera. The camera outputs a standard signal, displaying 30 frames per second, with each frame containing 350 lines. If the lens's field of view is 45mm, to ensure a measurement accuracy of 0.2mm (i.e., 250 points per line, 0.4% resolution), and to accurately determine the gap boundary, 500 points per line are sampled. Therefore, the sampling frequency is: f = 30 + 350 + 500 = 5.25 (MHz). (Figures 3 and 4 are shown). Many similar schemes exist, all featuring front-end ADC sampling and back-end processing using a high-speed, high-performance CPU, placing high demands on both the ADC and CPU speeds. For non-contact object displacement detection projects, the actual measurement required is the width of the gap signal. The video signal is processed to separate frame synchronization, line synchronization, and line image signals. ANDing the gap's line image signal with a high-frequency clock pulse yields a composite signal. The number of high-frequency clock pulses in the composite signal is proportional to the gap width, allowing the gap width to be measured by calculating the number of high-frequency signal cycles. The principle is illustrated in Figure 4. The gap width is measured by ANDing the image analog signal of the gap with a high-frequency periodic signal and calculating the number of high-frequency signal cycles in the synthesized signal. Data processing is performed by an MCU in the backend. The temperature drift problem of the high-frequency oscillation signal is compensated for by applying the inherent functional relationship between the oscillation clock signal and temperature, which is a trend in sensor development. The CAN bus has the characteristics of high transmission rate, high data efficiency, and high real-time performance. We implemented sensor communication via CAN bus using the MCP2510. System software flowchart See Figures 5 and 6. Figures 5 and 6 show how this sensor is used in an engineering testing device to achieve non-contact measurement of the gap displacement of a scale. Field testing showed that the sensor's non-contact measurement range is 0–45 mm, the maximum error is less than 0.2 mm, the response time is less than 70 ms, and the sensor cost is only a few hundred yuan.