Technological developments

Digital image processing technologies These technologies are used in our products and are available for licensing: (click the links to get more information) Super-resolution Spatial Resolution enhancement for both still images and video, number one solution in quality and speed. Image registration Precise image registration with sub-pixel precision and compensation for geometry distortions Radiometric alignment Matching exposure curves of differently exposed images with compensation for non uniform lighting changes Image fusion Expansion of Dynamic Range by fusing differently exposed images Expansion of Depth of Field by fusing differently focused images Detecting moving objects in a set of images, removing those objects or taking them from certain image Automatic correction of chromatic aberrations The only solution for entirely automatic correction of chromatic aberrations. Fixing oversharpening artefacts Fixing oversharpening artefacts after any type of image processing Seamless image segmentation and stitching Fast and efficient graph partitioning method of image segmentation Blurless Exposure Drastically reduces the amount of blur in mobile photos Superzoom A replacement for digital zoom, providing much higher quality zoomed photography on mobile devices

Licensing The technologies developed by Almalence, Inc. are available for licensing as binary-form SDKs and as source code as well. We offer very flexible licensing terms suitable for a wide range of applications – from custom in-house software to products marketed worldwide. For a licensing inquiry, please fill the form at the bottom of this page.

Super-resolution Super-resolution is a method for enhancing the resolution of an image or of an optical system. This can be done by either acquiring more graphic information (e.g. taking and merging several images of the same object) or by reconstructing details lost due to the faults of the optical system (applying knowledge of the optical system's properties). Our technology uses both ways to achieve maximum image quality. Almalence's super-resolution technology supports automatic algorithm tuning to particular cameras and optics. That makes it possible to achieve maximal image improvement for every imaging system. Unlike many other existing super-resolution methods, our method is non-iterative. With this method, it is possible to process even 20-30 megapixel images on an ordinary desktop computer within minutes. Other methods might require enormous amounts of memory and hours of time for calculations. Moreover, our method provides the best output quality compared to other methods. The super-resolution module of PhotoAcute SDK, which is available for licensing, contains super-resolution processing functions and a number of camera/lens profiles for super-resolution processing. We also deliver the tool for creation of profiles. Small-footprint adaptation of super-resolution technology The small-footprint adaptation of our technology, used in PhotoAcute Mobile application, combines high quality results with low CPU and memory usage, so that it can be used in devices with limited resource. The unique algorithms developed by our team solve super-resolution problems within the tight resource requirements of mobile devices where CPU and memory resources are severely limited. Read more: Superresolution FAQ Superresolution testing by MTF measurement Superresolution methods comparison Still image superresolution examples Superresolution video example Comparison of various superresolution methods with examples

Image registration Almalence's precise image registration algorithms allow: registering the images with sub-pixel precision; registering with compensation for non-linear geometry distortions, both regular (lens distortions) and irregular (such as atmospheric disturbances); registering images containing moving objects, taken at different focus points and/or with different exposures and changing lighting; using information about sensor structure for better registration of RAW images. You will be able to add this functionality to your applications by using PhotoAcute Elastic Image Aligner module of PhotoAcute SDK.

Precise radiometric alignment A precise radiometric alignment is desirable in image fusion tasks especially when a number of fusion techniques are applied in parallel, e.g. when merging a stack of exposure-bracketed images and concurrently processing the scene to remove duplicates of moving objects. Read more: Precise radiometric alignment. Producing consistent HDR images of scenes with moving objects and varying illumination.

Image fusion The Image Fusion module from PhotoAcute SDK is available for licensing and provides the following functionality: Expansion of dynamic range, Expansion of depth of field, Handling moving objects - removed completely or leaving only those present in a selected image. The Image Fusion module of PhotoAcute SDK provides "Noise-free HDR" - a technique of dynamic range expansion which makes it possible to reduce the noise in HDR images, especially in shadows. The same functionality can be used in normal (non-HDR) photography to eliminate noise even from the deep shadows of images by blending normally exposed and overexposed photos of the same object. Read more: Noise-free HDR: Special method of dynamic range expansion allows to remove the noise from HDR images

Correction of Chromatic Aberrations To a greater or lesser extent, every lens focuses the light of different wavelengths onto different focal planes and magnifies it differently. Correspondingly, this causes so-called "Longitudinal Chromatic Aberrations" and "Lateral Chromatic Aberrations". Chromatic aberrations are visible as the fringes of color at the contrast edges. The Chromatic Aberrations Correction module, which is available for licensing, provides functions which automatically correct aberrations, minimizing the circle of confusion. It is the only available technology for completely automatic CA correction. Read more: Chromatic aberration correction demo

Over-sharpening correction Some image processing algorithms happen to introduce over-sharpening artefacts to the image. A proprietary patented algorithm developed by Almalence automatically detects the level of over-sharpening in the image and corrects it. Only the over-sharpened areas/edges are corrected; the algorithm does not bring blurriness to the image areas unaffected by over-sharpening.

Seamless image segmentation and stitching Many modern image processing algorithms require a fast and efficient method of image segmentation. The algorithm we present achieves results similar to the well known graph-cut but using only a fraction of the time. Image processing algorithms such as panorama stitching, HDR fusion of dynamic scenes, or moving objects removal will benefit.

Blurless exposure Blurry and noisy images are a well-known problem of mobile imaging devices when shooting under less-then-perfect lighting conditions. The blurless exposure technique drastically reduces the amount of blur in images (both due to hand-shaking and moving objects) while retaining a low noise level and slightly extending the dynamic range. Read more: Blurless Exposure whitepaper (pdf) Real-life demo application for Nokia N900 Blurless Exposure examples

Superzoom Superzoom technology is the combination of super-resolution technology adapted to mobile platforms and a modification to the image sensor driver so as to obtain multiple sensor readouts within a fraction of a second. Unlike full-blown super-resolution processing, this technique does not require the user to take several continuous photos and takes mere seconds to provide the user with a high-quality result. This technique can be used as a drop-in replacement for digital zoom, providing much higher quality zoomed photography on mobile devices.   Source: http://almalence.com/technologies.html