company blog about SONY IMX637-AAMR-C Diagonal 3.9mm Event-Based Vision Sensor

Shenzhen Mingjiada Electronics Co., Ltd. supplies the Sony IMX637-AAMR-C, a 3.9 mm diagonal event-driven image sensor.

The Sony IMX637-AAMR-C is a 3.9 mm (1/4.5-inch) stacked event-driven vision sensor (EVS) jointly developed by Sony Semiconductor and Prophesee. It is designed for lightweight, low-latency and high-dynamic-range performance, overcoming the limitations of fixed-frame-rate sampling inherent in traditional frame-based image sensors. Centred on pixel-level asynchronous sensing, the sensor captures only information regarding changes in image brightness, discarding redundant static image data. It balances compact dimensions, high-speed response and efficient data output, making it widely suitable for high-speed dynamic vision scenarios such as industrial automation, intelligent robotics, security surveillance and precision measurement. It serves as the core hardware solution for high-precision dynamic vision capture in small to medium-sized applications.

I. IMX637-AAMR-C Product Positioning and Hardware Specifications

The IMX637-AAMR-C employs an industry-leading stacked process architecture with an optical format of 3.9 mm diagonal (Type 1/4.5). As a lightweight industrial-grade sensor that balances compact size with imaging coverage, it is perfectly suited to the integration requirements of miniaturised devices. Its core hardware parameters are precisely tailored to high-speed dynamic vision capture scenarios, with key specifications as follows:

- Pixel Specifications: Effective pixels approximately 0.32MP, resolution 640(H) × 512(V), full-pixel active mode with no redundant dead pixels; pixel size 4.86μm × 4.86μm, representing one of the smallest pixel sizes among sensors of this type, further reducing the overall sensor footprint

- Packaging and Power Supply: Utilises a 156-pin LGA package measuring 13mm × 13mm, suitable for compact device layouts; features a three-channel power supply architecture (analogue 3.0V, digital 1.1V, auxiliary 1.8V), ensuring stable power delivery and controllable power consumption to meet the demands of prolonged continuous operation

- Transmission Interfaces: Supports dual-mode high-speed output; the MIPI interface allows switching between 1 and 2 channels, whilst the SLVS interface supports switching between 2 and 4 channels. It is compatible with mainstream industrial vision transmission protocols and adapts to device systems with varying bandwidths

- Data formats: Compatible with the two major mainstream event output formats, EVT 2.1 and EVT 3.0, supporting 64-bit basic output and 16-bit high-efficiency encoded output respectively, enabling compressed transmission of event data to reduce system bandwidth demands

II. IMX637-AAMR-C Core Operating Principle: Event-Driven Asynchronous Detection

Unlike the operating mode of traditional CMOS sensors, which output full frames at a fixed frame rate, the IMX637-AAMR-C employs a pixel-level asynchronous event detection mechanism. Each pixel can independently monitor changes in image brightness in real time, without the need for global timing synchronisation. The sensor incorporates a built-in logarithmic I/V conversion circuit, an illuminance change detection unit and an arbitration circuit, enabling it to autonomously determine whether pixel brightness changes exceed preset thresholds.

When a pixel detects a valid increase or decrease in brightness, it immediately outputs the pixel’s coordinates (X/Y), a high-precision timestamp and information regarding the polarity of the brightness change in an asynchronous manner; for static pixels with no brightness changes, no data is output at any stage. This operating mode eliminates vast amounts of invalid static image data at source, realising an efficient acquisition logic of ‘capturing only dynamic changes and zero redundancy for static images’, thereby significantly enhancing the efficiency of visual data processing.

III. IMX637-AAMR-C Core Performance Advantages

1. Ultra-low latency, high-speed dynamic capture

The sensor boasts industry-leading low-latency performance. Under standard 1000 lx illumination, the maximum latency in a standard 9×9 pixel detection scenario is just 100 μs, with peak latency as low as 50 μs; even in low-light conditions of 5 lx, maximum latency can be kept within 1000 μs, whilst 1/5 subsampling mode is supported to further optimise latency performance. Its ultra-fast response speed enables precise capture of instantaneous changes in high-speed moving targets, eliminating the motion blur and lag issues associated with traditional frame sensors, making it suitable for applications such as high-speed motion tracking and instantaneous fault detection.

2. Ultra-high dynamic range, suitable for all scenarios

Thanks to the built-in logarithmic photoconversion circuit, the sensor’s output voltage is logarithmically correlated with incident light intensity, achieving an ultra-high dynamic range of 110dB. This enables the simultaneous capture of detail in both bright, high-light areas and dim, shadowed areas. Whether in direct outdoor sunlight, indoor environments with alternating light and shadow, or backlit workshops and low-light working conditions, it can accurately identify dynamic changes in the image without issues of overexposure or underexposure distortion, demonstrating exceptional environmental adaptability.

3. High-precision anti-interference, stable and reliable output

The sensor is equipped with multiple proprietary calibration and anti-interference functions. Under standard operating conditions, the typical contrast threshold is 25%, with threshold non-uniformity ≤6%, ensuring extremely high pixel detection accuracy. It also integrates functions such as analogue noise reduction filtering, anti-flicker, event filtering and dead zone time adjustment. These effectively suppress false event outputs caused by environmental light flickering and circuit noise. Under standard illumination of 1000 lx, the background noise frame rate is as low as 0.1 Hz, significantly reducing interference from invalid data and ensuring the accuracy of dynamic event recognition.

4. Multi-functional and adjustable, suitable for customisation across multiple scenarios

The device supports comprehensive customisation of parameters, including core settings such as contrast threshold, delay duration and dead time, allowing detection sensitivity to be precisely tailored to specific scenario requirements. It also features advanced functions such as pixel ROI cropping, digital pixel masking, event polarity inversion, multi-sensor master-slave synchronisation and external triggering. With a built-in thermometer and lux meter, it can monitor the operating environment and lighting conditions in real time, meeting the requirements for industrial customisation and multi-device coordination.

IV. Key Technical Highlights of the IMX637-AAMR-C

The IMX637-AAMR-C achieves multiple technical breakthroughs by leveraging Sony’s mature stacked CMOS process and Prophesee’s event-driven algorithms. Firstly, the ultra-small 4.86μm pixel size ensures detection accuracy whilst drastically reducing the sensor’s footprint, making it the preferred choice for compact smart devices; Secondly, the asynchronous event-driven architecture completely eliminates frame rate limitations, offering a temporal resolution far surpassing that of traditional frame cameras, capable of capturing dynamic changes at the millisecond and microsecond levels; Thirdly, the lightweight data output mode significantly reduces the computational load on the backend chip, eliminating the need to process vast amounts of static pixel data, thereby lowering device power consumption and hardware costs.

Furthermore, the sensor supports multi-mode event-compressed output; the EVT 3.0 encoding format significantly reduces data volume, thereby lowering transmission bandwidth and storage resource consumption whilst ensuring the integrity of effective dynamic information, making it suitable for embedded devices with limited computational capabilities.

V. Main Application Scenarios for the IMX637-AAMR-C

Thanks to its comprehensive advantages of compact size, low latency, high stability and low power consumption, the IMX637-AAMR-C is ideally suited to various high-speed dynamic visual perception scenarios. Its core application areas are as follows:

- Industrial Automation: High-speed assembly line material inspection, dynamic trajectory tracking for robotic arms, micro-motion fault monitoring for precision equipment, and dynamic defect screening in industrial settings. It is suited to high-speed production environments, enhancing inspection accuracy and efficiency

- Intelligent Robotics: Real-time obstacle avoidance, dynamic environmental perception and path trajectory correction for service robots and industrial AGVs; rapid response to environmental changes, enhancing robotic motion stability

- Intelligent Security: All-weather dynamic intrusion detection and movement alerts in unmanned areas; filtering out interference from static images to accurately identify valid movement events, reducing false alarm rates; suitable for complex lighting conditions day and night

- Precision Measurement and Research: Microscopic dynamic displacement measurement, high-speed motion object trajectory recording, and high-precision sequential visual data acquisition in laboratories; ensures measurement accuracy through microsecond-level latency and high dynamic range

- Smart Interaction and Esports: Motion capture for motion-sensing devices; high-speed response and recognition for esports interaction devices, delivering a lag-free human-machine interaction experience

VI. IMX637-AAMR-C Product Summary

As a 3.9mm compact event-driven vision sensor, the Sony IMX637-AAMR-C perfectly balances four core advantages: compact size, microsecond-level low latency, ultra-high dynamic range, and low-power, high-efficiency data output. Compared to traditional frame-based image sensors, its operational logic—which discards irrelevant data and focuses on dynamic events—is better suited to modern demands for high-speed, real-time, and lightweight smart vision. Stable industrial-grade performance, extensive customisation options, and compatible expansion interfaces make it a core component for lightweight, high-speed vision capture in fields such as industrial automation, intelligent robotics, and high-end security, providing efficient and reliable hardware support for real-time dynamic vision perception solutions.