Patent No. US10171740 (titled "Method and apparatus to correct blur in all or part of a digital image by combining plurality of images") on Sep 25, 2017. The application was issued on Jan 1, 2019.
’740 is related to the field of digital image processing, specifically addressing the problem of motion blur. In traditional photography, relative movement between the camera and the subject during a long exposure results in a smudged or blurred image. This is particularly problematic in low-light conditions where slow shutter speeds are required to gather enough light for a usable exposure.
The underlying idea behind ’740 is to replace a single long exposure with a sequence of high-speed captures that are intelligently aligned and merged. By using a fast shutter speed for each individual frame, the system freezes motion to prevent blur within those frames. These sharp but potentially underexposed images are then computationally stacked to achieve the light-gathering benefits of a long exposure without the associated motion artifacts.
The claims of ’740 focus on a method and device that allow for selective blur correction based on a designated subject. The system identifies a specific object within a preview image—either automatically or via user input—and processes a plurality of captured images to ensure that this primary subject remains sharp in the final output. This is achieved by aligning the frames specifically to the movement of that subject, even if it results in the rest of the scene remaining blurred.
In practice, the invention functions by capturing images back-to-back at the fastest possible shutter speed the hardware allows. A processor then performs pixel-by-pixel alignment, shifting each frame in the series to compensate for the measured or detected displacement of the target subject. When these aligned frames are summed or averaged, the signal-to-noise ratio improves significantly, effectively simulating a longer exposure while maintaining high edge definition for the tracked object.
This approach differs from prior solutions, such as mechanical image stabilization, which physically move lens elements or sensors to counter camera shake. While mechanical systems only compensate for the photographer's hand movements, this invention can correct for the independent motion of subjects within the frame. Furthermore, by allowing a user to toggle between different subjects for correction, the system provides a level of post-capture compositional control that traditional hardware-based stabilization cannot offer.
In the mid-2000s when ’740 was filed, digital image capture was typically implemented using CCD or CMOS sensors that recorded light impressions continuously for the duration of a shutter cycle. At a time when systems commonly relied on mechanical shutter speeds to mitigate motion blur, hardware and software constraints made the recovery of high-frequency image data non-trivial once relative motion between the camera and subject had occurred. During this era, standard post-processing techniques were generally limited to contrast and sharpness adjustments that often resulted in permanent data loss rather than a reconstruction of the original scene, while mechanical stabilization was largely confined to expensive, heavy lens-based shifting elements.
The disclosed invention represents a technical advancement through an architectural shift from mechanical stabilization to digital deconvolution and sensor-level compensation. It addresses the problem of image degradation caused by uniform relative motion by modeling the blur as a two-dimensional transfer function. The solution integrates motion sensors or blind estimation techniques to derive this function, which is then inverted to create a deconvolution filter that reverses the blurring effect. This integration enables the recovery of a true image from a smudged representation, overcoming the technical constraint of data loss inherent in traditional sharpening algorithms and providing a lighter, more cost-effective alternative to electro-mechanical lens stabilization.
The patent contains a total of 29 claims, with claims 1, 2, 10, and 20 being independent. These independent claims focus on methods and imaging devices designed to capture and combine multiple images to create a final output where a designated subject remains blur-free while other subjects or areas are blurred. The dependent claims serve to provide additional technical details regarding pixel value calculations, user interface indicators, image alignment techniques, the number of images processed, and the sequential timing of image capture.
Definitions of key terms used in the patent claims.
US Latest litigation cases involving this patent.

The dossier documents provide a comprehensive record of the patent's prosecution history - including filings, correspondence, and decisions made by patent offices - and are crucial for understanding the patent's legal journey and any challenges it may have faced during examination.
Get instant alerts for new documents