Audited Supplier
This product is a fully domestically produced dual-channel (infrared/visible light) embedded image processing board, measuring 7cm x 7cm x 1.7cm. It consists mainly of a dual-channel image processing baseboard (which includes processor interfaces and video input/output interfaces) and an RK3588 main processor core module. The RK3588 core processor module is mounted on the baseboard, forming the complete dual-channel image processing board.
Equipped with multiple video and communication interfaces.
Supports AI target recognition, target tracking, electronic image stabilization, image enhancement, image fusion, image stitching, and scene matching functions.
Operating temperature: -40°C to +60°C.
Designed with electromagnetic compatibility.
Strict quality control.
| CPU | Rockchips RK3588,64bit,8nm |
| 4x ARM Cortex-A76 RK3588J basic Frequency:normal mode 1.6GHz,overdrive mode 2.0GHz RK3588 basic Frequency:2.4GHz | |
| 4x ARM Cortex-A55 RK3588J basic Frequency:normal mode 1.3GHz,overdrive mode 1.7GHz RK3588 basic:1.8GHz | |
| NPU:6TOPS Support INT4/INT8/INT16/FP16/BF16/TF32 Support TensorFlow/PyTorch/Caffe/MXNet deep learning frameworks | |
| GPU:Mali-G610 MP4,support OpenGL ES 1.1/2.0/3.2,OpenCL 2.2,Vulkan 1.2 | |
| ISP:2x ISP(ISP0/ISP1),suppor HDR,3DNR,support below input: 48M:8064x6048@15fps dual ISP 32M:6528x4898@30fps dual ISP 16M:4672x3504@30fps single ISP | |
| Decoder:support 8K@60fps H.265,8K@30fps H.264 video coding | |
| Encoder:support 8K@30fps H.265/H.264 video coding | |
| ROM | 128GByte eMMC |
| RAM | 8GByte LPDDR4X |
| LED | 1x power indicator |
| 2x User programmable indicator | |
| Video IN | 1x PAL |
| 1x SDI | |
| Video OUT | HDMI 2.1,support 8K@60fps SDI support 1080p@60fps max |
| Other hardware | 1x SD/MMC,support SD3.0,MMC4.51,4bit data bus width |
| 2x GMAC,Support RMII/RGMII PHY,10/100/1000Mbps self-adaption | |
| 3x USB3.1 Gen1 (USB3.1 HOST),high data rate | |
| 1x USB2.0 OTG | |
| 3x RS422 Support automatic flow control mode | |
| 30x Timer,64bit,Supports timed interrupts, 12 safe timers and 18 non safe timers | |
| 5x Watchdog,32-bit watchdog timer |
It can be mounted on various platforms such as drones, unmanned vehicles, sensors, and electro-optical systems. It has a wide range of applications, including visual navigation, scene matching, power inspection, environmental monitoring, disaster prevention and reduction, policer investigation evidence, key area security, perimeter surveillance, autonomous driving, smart city management, industrial production, and medical image analysis. Custom functionality is also available upon customer request.
AI Detection and Recognition
The self-developed AI detection and recognition algorithm, combined with optimized hardware design, reduces module size, weight, and power consumption. It employs advanced, fast, and accurate recognition algorithms, capable of identifying various types of targets with precision.
Image Enhancement
This system enhances visible light images by removing fog and improves infrared image detail, enhancing image quality and increasing the effective range for long-distance targets.
Image Fusion
Visible light and infrared video fusion technology geometrically registers images of the same scene captured by different electro-optical imaging sensors. An algorithm combines the advantages or complementary information from each image to create a new image. Multi-sensor image fusion significantly improves the recognition and detection capabilities of the electro-optical system, enhancing image quality and indirectly improving system performance in recognition and tracking.
Panoramic Stitching
Seamless panoramic stitching of multi-sensor images is crucial for panoramic electro-optical detection systems. The stitched images effectively expand the sensor's field of view, addressing the limitations of single-camera image resolution, field of view angle, and spatial coverage. For adjacent sensor images with overlapping regions, image stitching can generate wide-angle panoramic images (360 degrees, 2π hemispheres, 4π global panoramas) that simulate real scenes. Multi-sensor image stitching has broad applications in airborne, vehicle-mounted, and ship-borne panoramic electro-optical monitoring systems, expanding reconnaissance and situational awareness.
| Function | Technical specification | Parameters |
| Target Recognition | Type | Human,Vehicle,UAV etc |
| Accuracy rate | >=90% | |
| Min recognize size | 10*10 pixel | |
| Target tracking | function | Anti-occlusion, anti-scale, anti-rotation |
| Accuracy rate | 1 pixel | |
| Minimum Tracking Size | 3*3 pixel | |
| Image Enhancement | Visible light dehazing Enhancement | effective range increased by 30% |
| IR detail Enhancement | effective range increased by 30% | |
| Electronic Image Stabilization (EIS) | Mode | Translation, Rotation. |
| Accuracy | <=1pixel | |
| Image fusion | Waveband | IR/Visible light |
| Mode | Grey/Pseudo-Color Fusion | |
| Image Stitching | Video Channels | 4~8 channel v |
| Mode | 360-degree panorama / 4π spatial stitching. | |
| Effect | No distortion, seamless, no color difference. | |
| Scene Matching | Accuracy | <=2 pixel |
| Ground landmark capture probability | ³90% |
Xi'an Zhongke Lead IR-Tech Co., Ltd established in 2015, is a Research type enteprise based on intelligent electro-optical equipment and infrared thermal imaging technology, incubated by Xi'an Institute of Optics and Precision Mechanics of China Academy of Sciences.
Lead IR-Tech has three core centers: Photoelectric equipment integration, precision optical processing, and specialized lens calibration. They are equipped with over 400 advanced devices, including CNC polishing machines, photoelectric, coordinate measuring machines, interferometers, and photoelectric system testing systems. This enables LEAD IR-TECH to achieve the full process of design, processing, integration, and testing of photoelectric products.
