Abstract: A k-cluster residue number system includes a processor and a memory. The processor is used to generate a modular set composed of p coprime integers, generate a dynamic range by taking a product of the p coprime integers, generate row indices for all integers in the dynamic range, generate column indices for all integers in the dynamic range, and generate a look-up table according to the row indices, the column indices and all integers in the dynamic set. The memory is used to store the look-up table. The p coprime integers include 2.

Abstract: A system for operating a floating-to-fixed arithmetic framework includes a floating-to-fix arithmetic framework on an arithmetic operating hardware such as a central processing unit (CPU) for computing a floating pre-trained convolution neural network (CNN) model to a dynamic fixed-point CNN model. The dynamic fixed-point CNN model is capable of implementing a high performance convolution neural network (CNN) on a resource limited embedded system such as mobile phone or video cameras.

Abstract: An image processing system includes an image capturing device, a pixel binning device, a temporal filter, a first memory, a re-mosaic device, a second memory, and a blending device. The image capturing device is used for capturing a raw image. The pixel binning device is coupled to the image capturing device for outputting an enhanced image according to the raw image. The temporal filter is coupled to the pixel binning device for outputting a preview image according to the enhanced image. The first memory is used for buffering the raw image. The re-mosaic device is coupled to the first memory for outputting a processed image. The second memory is used for buffering the enhanced image. The blending device is coupled to the re-mosaic device and the second memory for outputting a snapshot image according to the processed image and the enhanced image.

Abstract: After inputting input data to a floating pre-trained convolution neural network to generate floating feature maps for each layer of the floating pre-trained CNN model, a statistical analysis on the floating feature maps is performed to generate a dynamic quantization range for each layer of the floating pre-trained CNN model. Based on the obtained quantization range for each layer, the proposed quantization methodologies quantize the floating pre-trained CNN model to generate the scalar factor of each layer and the fractional bit-width of a quantized CNN model. It enables the inference engine performs low-precision fixed-point arithmetic operations to generate a fixed-point inferred CNN model.

Abstract: A system for operating a floating-to-fixed arithmetic framework includes a floating-to-fix arithmetic framework on an arithmetic operating hardware such as a central processing unit (CPU) for computing a floating pre-trained convolution neural network (CNN) model to a dynamic fixed-point CNN model. The dynamic fixed-point CNN model is capable of implementing a high performance convolution neural network (CNN) on a resource limited embedded system such as mobile phone or video cameras.

Abstract: A face recognition module includes a near infrared flash, a master near infrared camera, an artificial intelligence NIR image model, an artificial intelligence original image model, and an artificial intelligence fusion model. The NIR flash flashes near infrared light. The master near infrared camera captures a NIR image. The artificial intelligence NIR image model processes the NIR image to generate NIR features. The artificial intelligence original image model processes a 2 dimensional second camera image to generate face features or color features. The artificial intelligence fusion model generates 3 dimensional face features, a depth map and an object's 3 dimensional model according to the NIR features, the face features and the color features.

Abstract: A monoscopic low-power mobile device is capable of creating real-time images and videos for multiple viewpoints from a single captured view. The device uses statistics from an autofocusing process to create a block depth map of a single capture view. Artifacts in the block depth map are reduced and an image depth map is created. Alternate views are created from the image depth map using a Z-buffer based surface recover process and a disparity map which is a function of a geometric vision model.

Abstract: Implementations relate to systems and methods for real-time image recognition and mobile visual searching. A mobile device, such as a cellular phone, acquires an image and pre-processes the acquired image to generate a visual search query based on objects detected in the acquired image. The visual search query includes the acquired image or a query image extracted therefrom and metadata associated with the detected objects. The mobile device wirelessly communicates the visual search query to a remote server, and in response to the visual search query, the remote server recognizes an object in the query image based on the associated metadata. The remote server then generates information content based on the recognized object and communicates the information content to the mobile device to be presented via the mobile device.

Abstract: This disclosure describes techniques for rotating an encoded image, such as an image encoded according to a JPEG standard. In one example, a method for rotating an encoded image comprising reordering minimum coded units (MCUs) of the encoded image according to a specified rotation of the encoded image, rotating image data within the MCUs according to the specified rotation, and generating a rotated version of the encoded image comprising the reordered MCUs and the rotated image data within the MCUs.

Abstract: A monoscopic low-power mobile device is capable of creating real-time stereo images and videos from a single captured view. The device uses statistics from an autofocusing process to create a block depth map of a single capture view. Artifacts in the block depth map are reduced and an image depth map is created. Stereo three-dimensional (3D) left and right views are created from the image depth map using a Z-buffer based 3D surface recover process and a disparity map which is a function of the geometry of binocular vision.

Abstract: A monoscopic low-power mobile device is capable of creating real-time images and videos for multiple viewpoints from a single captured view. The device uses statistics from an autofocusing process to create a block depth map of a single capture view. Artifacts in the block depth map are reduced and an image depth map is created. Alternate views are created from the image depth map using a Z-buffer based surface recover process and a disparity map which is a function of a geometric vision model.

Abstract: An apparatus is described. The apparatus includes a digital camera that is capable of shooting at least two images at different effective lens focal lengths. The camera is also capable of high pass filtering (HP) said at least two images, estimating respective spectral densities (SD) of said at least two filtered images and then low pass filtering (LP) the respective estimates of said at least two filtered images prior to combining said at least two images into a single image to produce one or more optical effects.

Abstract: Systems and methods to selectively combine images are disclosed. In a particular embodiment, an apparatus includes a registration circuit configured to generate a set of motion vector data based on first image data corresponding to a first image and second image data corresponding to a second image. The apparatus includes a combination circuit to selectively combine the first image data and adjusted second image data that corresponds to the second image data adjusted according to the motion vector data. The apparatus further includes a control circuit to control the combination circuit to generate third image data.

Abstract: To combine two or more images, an image capture device may compute motion vectors for a plurality of blocks of pixels of one of the images. The image capture device may also interpolate or extrapolate motion vectors for individual pixels or sub-blocks of pixels using the block motion vectors. The image capture device may then average the first and second images by averaging each of the pixels of the first image with pixels of the second image that correspond to a location indicated by the plurality of motion vectors. These techniques may reduce blur in image information resulting from certain movements during image capture or the use of certain image capture technologies.

Abstract: This disclosure describes barcode scanning techniques for an image capture device. The image capture device may automatically detect a barcode within an image while the image capture device is operating in a non-barcode image capture mode, such a default image capture mode. In one aspect, the detection of the barcode within the image may be based on a combination of identified edges and low intensity regions within the image. The image capture device may configure, based on the detection of the barcode, one or more image capture properties associated with the image capture device to improve a quality at which the images are captured. The image capture device captures the image in accordance with the configured image capture properties. The techniques may effectively provide a universal and integrated front-end for producing improved quality images of barcodes without requiring significant interaction with a user via a complicated user interface.

Abstract: Systems and methods to selectively combine images are disclosed. In a particular embodiment, an apparatus includes a registration circuit configured to generate a set of motion vector data based on first image data corresponding to a first image and second image data corresponding to a second image. The apparatus includes a combination circuit to selectively combine the first image data and adjusted second image data that corresponds to the second image data adjusted according to the motion vector data. The apparatus further includes a control circuit to control the combination circuit to generate third image data.

Abstract: A method and apparatus for generating stereoscopic images of a scene is described. The apparatus may have a first image sensor, a second image sensor spaced apart from the first image sensor, a diversity combine module to combine image data from the first and second image sensors, and an image processing module configured to process combined image data from the diversity combine module may be used to generate stereoscopic images of a scene.

Abstract: A device has a processing unit to implement a set of operations to use both luma and chroma information from a scene of an image to dynamically adjust exposure time and sensor gain. The processing unit collects bright near grey pixels and high chroma pixels in the scene. Based on the collected pixels, brightness of the near grey pixels is increased to a predetermined level without saturation. At the same time, the high chroma pixels are kept away from saturation.

Abstract: This disclosure describes techniques for detecting a barcode within an image. An image processor may, for example, process an image to detect regions within the image that may be barcodes. The image processor may identify regions of the image that exhibit a high concentration of edges and a high concentration of pixels with low optical intensity co-instantaneously as potential barcodes. The image processor may identify the regions using a number of morphological operations. The image processor may then determine whether the identified regions are actually barcodes by verifying whether the region have unique barcode features. The barcode detection techniques described in this disclosure may be independent of barcode size, location and orientation within the image. Moreover, the use of morphological operations results in faster and more computationally efficient barcode detection, as well as lower computational complexity.

Abstract: Systems and methods of high dynamic range image combining are disclosed. In a particular embodiment, a device includes a global mapping module configured to generate first globally mapped luminance values within a region of an image, a local mapping module configured to generate second locally mapped luminance values within the region of the image, and a combination module configured to determine luminance values within a corresponding region of an output image using a weighted sum of the first globally mapped luminance values and the second locally mapped luminance values. A weight of the weighted sum is at least partially based on a luminance variation within the region of the image.