Sony Semiconductor Solutions Group


In the field of digital imaging, "noise" was once an insurmountable barrier standing in the way of clear vision. It not only erodes the details of images but also restricts the perception capability of smart devices in complex environments. For instance, in low-light conditions, noise makes it difficult for security cameras to accurately capture key details in surveillance footage, thereby compromising the effectiveness of security monitoring. Autonomous vehicles are also affected: noise may cause sensors to fail in accurately judging road conditions at night or in severe weather, increasing driving risks.

Traditional noise reduction technologies often offer only a temporary fix rather than addressing the root cause. They merely suppress noise through algorithms after image generation, which leads to image distortion or loss of details. A typical example is that the accuracy of face recognition systems degrades due to noise, rendering them ineffective in complex environments.

However, with the in-depth research and development of Shanghai Dishan Technology Co., Ltd. in the field of semiconductor image sensor chip technology, this long-standing pain point of the industry is now being tackled with a groundbreaking solution. Guided by the technological philosophy of "striving for perfection", this company, driven by an unquenchable thirst for exploration, is reconstructing the logic of noise reduction from the underlying technology level, redefining the performance boundaries of image sensors, and bringing a revolutionary imaging experience to the entire industry.

Core Breakthrough: Suppress Noise at the Source and Redefine the Signal-to-Noise Ratio

Dishan Technology’s noise reduction philosophy is not a simple case of "post-facto remediation", but rather a commitment to intervening at the very source of photoelectric conversion.

After years of dedicated research and development, the company has made substantial strides in low-noise performance, successfully reducing the noise of image sensors by 30 dB. This remarkable figure has not only set a new industry record, but also signifies a qualitative leap in the sensor’s ability to capture weak light signals.

Behind this breakthrough lies Dishan Technology’s relentless push against physical limits:

Material and Architecture Innovation: Breaking Through Traditional Physical Boundaries

The R&D team has innovatively adopted advanced photoelectric conversion materials and technical architectures. This new technology, based on nanoscale semiconductor composites, not only drastically improves photoelectric conversion efficiency—boosting the light signal capture capacity per unit area to 2.5 times that of traditional materials—but also fundamentally reduces interference during signal transmission through its unique design leveraging the photon tunneling effect.

The photon tunneling effect is a quantum mechanical phenomenon that allows photons to pass through materials conventionally regarded as insulators, thereby significantly enhancing the efficiency and accuracy of signal transmission. In low-light conditions, this advantage translates into a 33% improvement in imaging quality and a 46% increase in color accuracy.

For example, in extremely dark environments with only 0.1 lux of illumination (equivalent to starlight conditions), devices equipped with Dishan Technology’s sensors can still deliver near daylight-level clarity and color reproduction. The resulting images are pristine and free of noise, as if a "noise elimination button" has been pressed for the night.

 

Circuit-Level Noise Reduction: Precision Control in the Microcosm

Noise often arises from subtle fluctuations in circuits. By optimizing circuit design, adopting custom low-noise transistors and multi-stage signal amplification technology, Dishan Technology has effectively suppressed read noise and dark current noise. The team has even managed to control circuit noise to below 0.1 e⁻ RMS (electron root mean square), an indicator that has reached the world’s top level.

This "circuit-level" purity control is akin to building a "signal highway" in the microcosm, enabling sensors to capture weak signals without being swamped by useless electronic interference, thus achieving a qualitative leap in image signal-to-noise ratio.

Benefiting from this technology, Dishan Technology’s products have achieved a breakthrough in the dynamic range (DR) indicator of 120 dB, far exceeding the industry average of 80 dB. This means the sensors can clearly capture scenes with extreme light intensity ranges simultaneously, from the glare of scorching sunlight to the faint glow of candle flames.

Ultimate Experience: Transcending the Boundaries of Light and Speed

Extending to the "microscopic" and "low-light" poles: see and not see

Dishan Technology's breakthroughs in CMOS image sensors and noise reduction technology have solved the bottleneck of traditional imaging in extreme environments, allowing machines to "see more clearly and realistically".

Revolution in low-light imaging: Traditional CMOS sensors tend to be noisy and blurry in low-light environments. By improving the internal structure of the sensor, using advanced materials, and combining the powerful computing power of its AI GPU, Dishan Technology has significantly improved the sensitivity of the sensor.

Experience upgrade: This allows the device to output clear, low-noise images even in extremely dark environments (such as night surveillance, underground pipe gallery detection), pushing the imaging experience from "visible" to "clear color and detail".

Industry impact: In the field of security monitoring, this means that even at 0.01Lux illumination, it can effectively suppress color noise, support face recognition and motion detection, and greatly improve the level of public safety.

Extreme sensor performance: The new sensor developed by the company not only improves image quality, but also reduces power consumption. This low power consumption and high sensitivity are the core requirements of smart wearable devices (such as AI glasses) and IoT terminals.

2. Extend to the "macro" and "ultra-high-definition" poles: see the whole picture

If micro is to solve the problem of "details", then macro is to solve the problem of "vision" and "capacity". Although Dishan Technology mainly provides core computing power and sensing components, its technology supports the realization of 100-million-level pixel imaging systems.

Global Perception of Billions of Pixels: In security and industrial inspection, traditional megapixel cameras turn into a mosaic when zooming in to see distant details. Dishan Technology's underlying technologies (such as 2nm process GPUs and high-performance image processing) provide the necessary computing power support for systems with billions of pixels (such as 1 billion or even 2.4 billion pixels).

Experience upgrade: This technology allows users to accurately capture sparrows with a wingspan of 20 cm from a few kilometers away, or monitor thousands of targets simultaneously on vast scenic spots and airport runways. It's an experiential leap from "single point of monitoring" to "global recording."

Industry impact:

Ecological protection: Accurately observe the migration of migratory birds and establish a biological characteristic bank.

Smart airports: Real-time tracking of aircraft, vehicles, and personnel on the runway to prevent collisions and improve operational efficiency.

Low-altitude economy: Monitor drone distribution lanes to provide "eyes" for future air traffic management.

The "polar" contrast of the imaging experience

To more intuitively understand this change "extending to the poles", here is a comparison table:

Dimensions

Micro/low-light extreme scenes

Macro/UHD extreme scenes

Core Challenge

Poor lighting, noise interference, loss of detail

Broad vision, small target, and huge amount of data

The supporting technology of Dishan Technology

New low-power CMOS sensor, AI noise reduction algorithm, 2nm process GPU

High-performance AI GPU computing power, heterogeneous computing architecture, image processing chips

Experience keywords

Clear, pure, and true

Panoramic, detailed, at the same time

Typical application scenarios

Night face recognition, autonomous driving night vision, medical endoscopes

Smart airport management, wildlife monitoring, smart city screen

 Revolutionary changes for the industry

Dishan Technology's technology is not just a pile of parameters, it is reconstructing the industry's operating mode:

From "people looking for information" to "information looking for people": Relying on hundreds of millions of pixels and intelligent analysis, the system can automatically identify tiny targets of interest (such as a specific bird or a violator) in the huge picture, and issue early warnings, greatly reducing the fatigue of manual screen staring.

Breaking Physical Limits: Through the combination of "optics + computing", it compensates for the limitations of the physical focal length and sensor size of traditional lenses. For example, on mobile phones or lightweight AI glasses, imaging effects close to professional equipment can also be achieved.

Empowering localization substitution: Dishan Technology has achieved complete autonomy and control of core IP, and has developed a tool chain compatible with mainstream ecology. This means that domestic manufacturers no longer rely entirely on imported high-end sensors and chips, and can carry out more in-depth customized development in security, autonomous driving and other fields.

Through the dual breakthroughs in sensors and computing power, Dishan Technology has indeed pushed the imaging experience to the extreme - not only allowing us to see every detail in the dark, but also allowing us to control every small change in the vast world. This is the best proof that technology serves "experience".

 Break Through the "Nighttime" Limitation and Unlock All-Weather Vision

Traditional image sensors often need to boost the ISO sensitivity to maintain image brightness in nighttime or low-light environments, but this usually comes with a dramatic surge in noise, creating a paradox where brightness and sharpness cannot be achieved simultaneously.

Dishan Technology’s technology has broken this vicious cycle. Its sensors, through the in-depth integration of hardware-level noise reduction and AI algorithms, can maintain high signal-to-noise ratio (SNR) output even under extremely low illumination.

For example, in nighttime urban surveillance scenarios, cameras equipped with this technology can not only clearly identify license plate numbers and pedestrians’ facial features, but even capture the subtle sway of tree branches a hundred meters away, providing the security field with an innovative solution that ensures clearer, farther, and more accurate imaging. In the field of astrophotography, its sensors can capture finer details of dim deep-space celestial bodies, revealing the beauty of the starry sky with stunning clarity. In lightless environments such as underground exploration, devices can also achieve precise detection via low-light imaging technology, facilitating resource development and scientific research.

Race Ahead on the "Speed" Track, Freeze Millisecond Moments

In high-speed motion scenarios—such as industrial assembly line inspection, high-speed drone flight, or sports event capture—traditional sensors often suffer from motion blur or the rolling shutter effect due to slow readout speeds, coupled with noise accumulation, which results in the loss of critical moments.

The innovative signal readout circuit developed by Dishan Technology adopts a parallel multi-channel transmission architecture and global shutter technology, boosting the image data transmission rate by 5 times to an impressive 40 Gbps, whereas the typical transmission rate of traditional sensors is only 8 Gbps. Combined with its low-noise characteristics, this technology ensures that every frame maintains ultra-high purity and sharpness during high-speed continuous shooting (tens of thousands of frames per second) or high-frame-rate recording (e.g., 8K 120fps video).

For instance, in motor racing events, cameras can clearly capture the transient deformation of racing car tires in contact with the ground—a feat difficult to achieve with traditional sensors. In the field of industrial quality inspection, tiny defects on high-speed production lines can also be accurately identified, improving defect detection accuracy by 30% compared with traditional technologies, and significantly enhancing production efficiency and product quality.

Empowering the Future: Enabling Machines to "See" with Greater Insight and Building the Cornerstone of Intelligent Perception

Dishan Technology’s pursuit extends beyond visible improvements in image quality perceived by the human eye, and focuses more on enhancing the "understanding" capability of machine vision.

In the fields of artificial intelligence and robotics, image sensors serve as the "eyes" of machines, and the quality of the data they output directly affects the decision-making accuracy of AI algorithms. The presence of noise can severely interfere with image feature extraction, leading to reduced recognition rates and increased misjudgment rates. By reducing noise, Dishan Technology has significantly improved the contrast and detail sharpness of images, providing higher-quality "raw materials" for subsequent AI inference.

Intelligent Driving: The "Clairvoyant Eye" for Safe Perception

In Advanced Driver Assistance Systems (ADAS), the accuracy and real-time performance of environmental perception are the key to ensuring driving safety. Dishan Technology’s low-noise sensors can identify roadside obstacles, traffic signs and pedestrians earlier and more accurately.

For instance, when driving in rainy, foggy weather or at night, the sensors can penetrate harsh environments to clearly perceive road condition information within a 200-meter range ahead, providing millisecond-level decision-making basis for autonomous driving systems. Their high dynamic range (HDR) characteristics can also effectively suppress glare, ensuring that the cameras still work properly even when exposed to high beams from oncoming vehicles, thus avoiding "visual blind spots".

Industry 4.0: The "Quality Inspection Maestro" of the Microcosm

 

From "human eye experience" to "machine vision"

In traditional industrial production lines, micron-scale defect detection often relies on the naked eye observation of skilled workers. This is not only inefficient, but also highly susceptible to fatigue and subjective factors. The technological breakthrough of Dishan Technology directly solves the pain point of "unclear" machines:

The physical basis of extremely low noise: Through innovative circuit design and material application, the noise is greatly reduced (data shows that it can be reduced by 30dB), which means that the signal captured by the sensor is almost entirely real image information, not a "snowflake" of interference.

Micron-level detail reproduction: With high pixel integration (5.3% increase in pixel count), the sensor can capture scratches as small as 0.01 mm on metal surfaces and micron-scale solder joint defects on circuit boards.

Revolutionary change: let invisible defects "nowhere to hide"

This improvement of technical capabilities directly broadens the "territory" of quality inspection, allowing it to cover more demanding scenarios:

Limitations of traditional quality inspection

Quality inspection capabilities empowered by Dishan Technology

Industry value

Difficult to identify transparent/reflective materials

Perspective on internal stress cracks: High signal-to-noise ratio combined with precision optics can penetrate the surface of transparent materials and detect tiny stress distributions and cracks inside.

Improve the safety of glass, optical lenses, and precision plastic parts.

Missed detection in low-light/shadow environments

Clear imaging in all scenes: Under the complex light of the production line, the picture is still kept pure and free of noise interference.

It solves the problem of low recognition rate of small defects in low-light environment.

Surface inspection only

Deep Defect Capture: Capable of catching tiny false solders or faults inside the board.

Greatly reduce the failure rate of electronic products after leaving the factory.

Technology Convergence: More Than Just the "Eye"

It is worth noting that Dishan Technology's business territory is not limited to image sensors. According to the data, Dishan Technology also has a deep layout in AI GPU and robotics.

This means that Dishan Technology provides not only a "high-definition camera", but a closed-loop system of "perception + computing":

Perception end: High signal-to-noise ratio sensor collects "pixel-level" pure images.

Processing end: Self-developed AI GPU (computing power up to 50 TFLOPS) for high-speed signal reading and parallel computing.

Decision-making side: AI algorithms quickly compare standard models and instantly determine defects.

The ultimate goal of technology is to serve the experience. In the industrial field, this "experience" is the certainty of production.

Through breakthroughs in noise reduction technology, with its accumulation in the field of AI computing power and robotics, Dishan Technology has pushed industrial quality inspection from "auxiliary tools" to the position of "core decision-makers". It allows machines to see details that humans cannot see, identify weak signals that humans cannot perceive, and truly push the manufacturing industry towards the ultimate goal of "zero defects".

Medical Imaging: The "Perspective Lens" for Life and Health
In the field of medical imaging, noise can obscure lesion details and compromise diagnostic accuracy. Dishan Technology’s low-noise technology provides a clearer imaging foundation for devices such as endoscopes, ultrasound probes, and CT scanners.For example, during endoscopic examinations, the sensor can render finer mucosal textures, helping doctors detect early signs of cancer at an early stage. In fetal ultrasound scans, its high signal-to-noise ratio (SNR) images can more accurately display fetal development conditions, laying a reliable foundation for medical diagnosis.
Unceasing Exploration Illuminates the Future of Technology

From a technical specification of 30dB noise reduction, to crystal-clear image performance in low-light conditions; from millisecond-level capture of high-speed scenes, to precise sensing that empowers the intelligent world, Dishan Technology has demonstrated its pursuit of "ultimate imaging" through tangible actions. This is not only a triumph of semiconductor materials and craftsmanship, but also a compelling testament to how Made-in-China Smart Manufacturing is breaking monopolies and leading innovations in the high-end sensor sector.

The company’s founder once stated: "There is no ceiling for technology, only endless stepping stones for breakthroughs. Our goal is not to 'be the best', but to 'keep getting better'."

In the future, with the boom of artificial intelligence, the metaverse and autonomous driving technologies, the performance requirements for image sensors will be boundless. Dishan Technology has already mapped out its next-generation technology roadmap, which includes ultra-high-sensitivity sensors based on quantum effects, in-depth integration of on-chip AI computing and imaging, and expanded sensing capabilities in the terahertz band.

Embracing the mindset of an "explorer", this enterprise is clearly poised to continue exploring the next technological singularity on the path of pursuing excellence. As the light of technology illuminates more uncharted territories, the name Dishan Technology is destined to become synonymous with "ultimate imaging", writing a new chapter in how humanity perceives the world.