FIXME **This page is not fully translated, yet. Please help completing the translation.**\\ // (remove this paragraph once the translation is finished) // ====Image Stabilization==== **1. What is Camera Shake**\\ When capturing a running child in a park with a smartphone, a blurred photo misses the lively moment; when operating a drone to photograph landscapes, severe camera shake in the footage ruins the charm of the magnificent scenery; at critical moments in security surveillance, blurred images make it impossible to capture important information... All these issues that affect the user experience point to the same core problem — **camera shake**.\\ Camera shake refers to the unintended displacement or vibration of a camera (or imaging device) during shooting caused by external disturbances. This instability directly causes the imaging to deviate from the ideal state: still photos will exhibit "motion blur" and "blurred details", while dynamic videos will show "vertical/horizontal jitter" and "inter-frame misalignment", ultimately severely affecting the viewing experience and the effectiveness of information delivery.\\ {{ yanding:抖动3.jpg?600 |}} **2. Image Stabilization Technologies**\\ The essence of camera image stabilization technology is to counteract the displacement caused by shake through hardware or algorithms. The current mainstream technologies include:\\ **1. Optical Image Stabilization (OIS)**\\ Principle: Detects device shake via a gyroscope and drives the lens or sensor to move horizontally/vertically to compensate for optical path deviation.\\ Advantages: No loss in image quality (no need to crop the image), especially suitable for telephoto and low-light scenarios.\\ **2. Electronic Image Stabilization (EIS)**\\ Principle: Based on software algorithms, it performs "pixel-level correction" for displacement caused by shake by cropping redundant pixels at the edges of the frame.\\ Advantages: Low cost (no additional hardware required), suitable for dynamic video stabilization; some models support an "enhanced mode" that combines gyroscope data to improve accuracy.\\ Limitations: Crops the image (typically losing 5%-15% of the field of view), which may lead to a decrease in image quality in extreme cases.\\ Although the aforementioned image stabilization technologies can largely suppress camera shake, they cannot completely eliminate it due to limitations in technical principles, shake amplitude, and other factors. The stabilization performance varies across different devices, and shake conditions in actual usage scenarios are complex and variable. Therefore, it is necessary to measure the image stabilization performance of cameras to accurately understand their actual performance in different scenarios. **3. How to Measure Camera Image Stabilization Performance**\\ The core objective of image stabilization performance measurement is to evaluate the performance of video and still cameras in suppressing shake and maintaining imaging stability. To achieve this goal, two key phenomena need to be measured: first, the motion blur caused by shake in video frames; and second, the image distortion that may occur during the video stabilization process.\\ **3.1 Test Standards:**\\ Image stabilization testing is based on the international standard ISO 20954-1:2019. **3.2 Test Equipment:**\\ Vertical multi-color temperature fill light source ([[https://www.yanding.com/product/detail?id=1534|LS-CCXL-2S06-IR]]), test chart ([[https://rdbuy.com/product/productdetail?model_id=10336|CP316 Reflective Gray Dead Leaves Chart]]), 2-axis stabilization kit (IS-Shake02)\\ {{ yanding:防抖00.png?800 |}} |{{ yanding:防抖设备.png?300 |}}| ^ IS-Shaker02 2-Axis Shake Platform ^ The IS-Shaker02 image stabilization test system is a high-precision test equipment independently developed by Yanding, specifically designed for evaluating camera image stabilization performance. It can simulate real-world shake scenarios through high-precision dual-axis vibration, and is applied to the image stabilization performance evaluation of devices such as smartphones, smart glasses, security cameras, and DSLR cameras. **3.3 Test Procedures:**\\ **1. Environment Setup:**\\ Conduct the test in a darkroom with an indoor illuminance of ≤2 lux; the illuminance uniformity on the surface of the test chart should be ≥90%;\\ Lighting environment: 6500K, 3000 lux; 6500K, 800 lux; 6500K, 30 lux;\\ Shake platform set frequencies and angles: 2Hz, 4°; 4Hz, 2.5°; 6Hz, 1.5°; 8Hz, 1°; 10Hz, 0.8°; 12Hz, 0.15°; 15Hz, 0.15°.\\ **2. Sample Acquisition**\\ The DUT must be fixed to the 2-axis stabilization platform so that the geometric center of the DUT coincides as much as possible with the rotation center of the platform, ensuring that its optical axis is as perpendicular as possible to the plane of the test chart. The test chart should occupy about one-third of the image height. After the image stabilizes and focus is clear, start the 2-axis stabilization platform and record video for a duration of no less than 15 seconds.\\ {{ yanding:场景照.png?600 |}} **3. RIQA Analysis**\\ Open the RIQA software, click on the chart directory, and select Shaking in the Camera video module. Click "+Add" to add the video to be analyzed, and set the frame interval. Enter 1 if every frame needs to be analyzed, or enter an appropriate frame interval, such as 10, to reduce the analysis time.\\ {{ yanding:防抖01.png?600 |}} Click the "Start" button, and the software will automatically analyze and output the data metric results of the recorded shake video. Click the "Generate Report" button to generate a detailed test report containing key metrics, chart comparisons, and clear conclusions with one click. {{ yanding:防抖02.png?600 |}} **4. Result Interpretation**\\ {{ yanding:防抖03.png?600 |}} The data metric results are described as follows:\\ ^ Metric ^ Definition ^ | Mean MTF50P | The average of half the MTF peak value across all frame data | | Max MTF50P | The maximum of half the MTF peak value across all frame data | | Mean Rise1090 | The average number of pixels occupied by the slanted edge grayscale from 10% to 90% across all frame data | | Max Rise1090 | The maximum number of pixels occupied by the slanted edge grayscale from 10% to 90% across all frame data | | Zoom(Maximum) | The maximum zoom ratio across all frame data | | Horizontal(Maximum) | The maximum horizontal distortion across all frame data | | Vertical(Maximum) | The maximum vertical distortion across all frame data | | Shear(Maximum) | The maximum shear angle across all frame data | | Verirical stretch(Maximum) | The maximum vertical stretch across all frame data | | Zoom(Std Dev) | The standard deviation of the zoom ratio across all frame data | | Horizontal(Std Dev) | The standard deviation of horizontal distortion across all frame data | | Vertical(Std Dev) | The standard deviation of vertical distortion across all frame data | | Shear(Std Dev) | The standard deviation of the shear angle across all frame data | | Verirical stretch(Std Dev) | The standard deviation of vertical stretch across all frame data | **See More**\\ [[:camera测试用例]], [[:测试用例]], [[:riqa简介]]