Abstract: An image processing circuit is provided. A memory control unit first moves a reference image and parameters from a first memory to a third memory. A convolution neural network circuit obtains the parameters from the third memory. The memory control unit also moves at least one portion of the reference image from the third memory to a second memory, in which a storage capacity of the third memory is larger than that of the second memory. A depth decoder obtains the reference image from the second memory for calculating depth information according to the reference image and a structured image received from an infrared sensor, and stores the depth information into the second memory.

Abstract: A framebuffer-less system of convolutional neural network (CNN) includes a region of interest (ROI) unit that extracts features, according to which a region of interest in an input image frame is generated; a convolutional neural network (CNN) unit that processes the region of interest of the input image frame to detect an object; and a tracking unit that compares the features extracted at different times, according to which the CNN unit selectively processes the input image frame.

Abstract: An image processing circuit is provided. A memory control unit first moves a reference image and parameters from a first memory to a third memory. A convolution neural network circuit obtains the parameters from the third memory. The memory control unit also moves at least one portion of the reference image from the third memory to a second memory, in which a storage capacity of the third memory is larger than that of the second memory. A depth decoder obtains the reference image from the second memory for calculating depth information according to the reference image and a structured image received from an infrared sensor, and stores the depth information into the second memory.

Abstract: A framebuffer-less system of convolutional neural network (CNN) includes a region of interest (ROI) unit that extracts features, according to which a region of interest in an input image frame is generated; a convolutional neural network (CNN) unit that processes the region of interest of the input image frame to detect an object; and a tracking unit that compares the features extracted at different times, according to which the CNN unit selectively processes the input image frame.

Abstract: A system of selecting a keyframe for iterative closest point (ICP) includes a reference frame selector that generates a reference frame according to a current frame and a current keyframe; an ICP loop unit that performs ICP on the reference frame and the current frame, thereby generating a pose of the current frame; and a keyframe update unit that generates a new keyframe according to an offset condition between the pose of the current frame and a pose of the reference frame.

Abstract: The present invention is directed to an adaptive method for object detection. A predetermined number of next window images following a current window image are skipped, if a current likelihood value is less than a predetermined background threshold. The object detection early terminates, if a previous window image preceding the current window image contains the object to be detected and the current likelihood value is greater than or equal to a predetermined foreground threshold.

Abstract: A system of selecting a keyframe for iterative closest point (ICP) includes a reference frame selector that generates a reference frame according to a current frame and a current keyframe; an ICP loop unit that performs ICP on the reference frame and the current frame, thereby generating a pose of the current frame; and a keyframe update unit that generates a new keyframe according to an offset condition between the pose of the current frame and a pose of the reference frame.

Abstract: A clustering method with a two-stage local binary pattern includes generating a gradient direction value according to a center sub-block and neighbor sub-blocks of a patch of an image; quantizing the gradient direction value, thereby generating a quantized gradient direction value; generating a gradient magnitude value according to the gradient direction value; quantizing the gradient magnitude value, thereby generating a quantized gradient magnitude value; concatenating the quantized gradient direction value and the quantized gradient magnitude value to generate a two-stage local binary pattern (2SLBP) value; and performing clustering of super-resolution imaging by using the 2SLBP value as an index.

Abstract: An image processing apparatus and an image processing method thereof are provided. A shared storage unit of a motion estimation and motion compensation apparatus captures frame data of a storage unit through a bus. A motion vector estimation unit and a motion compensation unit capture image data for executing a motion vector estimation operation and a motion compensation operation from the sharing storage unit.

Abstract: An iterative matching method for partial fingerprint verification includes providing database features and input features; initially comparing the input features with the database features using one of the database features as a first reference point, resulting in initial matched feature pairs between initial matched database features and corresponding initial matched input features; and progressively comparing the input features with the database features using a gravity center of the initial matched database features as a second reference point, resulting in progressive matched feature pairs between progressive matched database features and corresponding progressive matched input features.

Abstract: A clustering method with a two-stage local binary pattern includes generating a gradient direction value according to a center sub-block and neighbor sub-blocks of a patch of an image; quantizing the gradient direction value, thereby generating a quantized gradient direction value; generating a gradient magnitude value according to the gradient direction value; quantizing the gradient magnitude value, thereby generating a quantized gradient magnitude value; concatenating the quantized gradient direction value and the quantized gradient magnitude value to generate a two-stage local binary pattern (2SLBP) value; and performing clustering of super-resolution imaging by using the 2SLBP value as an index.

Abstract: An iterative matching method for partial fingerprint verification includes providing database features and input features; initially comparing the input features with the database features using one of the database features as a first reference point, resulting in initial matched feature pairs between initial matched database features and corresponding initial matched input features; and progressively comparing the input features with the database features using a gravity center of the initial matched database features as a second reference point, resulting in progressive matched feature pairs between progressive matched database features and corresponding progressive matched input features.

Abstract: An image processing apparatus and an image processing method thereof are provided. A shared storage unit of a motion estimation and motion compensation apparatus captures frame data of a storage unit through a bus. A motion vector estimation unit and a motion compensation unit capture image data for executing a motion vector estimation operation and a motion compensation operation from the sharing storage unit.

Abstract: A content adaptive compression system includes a plurality of encoders being coupled to receive a portion of an image, and accordingly generating candidate compressed codes, respectively, the plurality of encoders being configured for encoding images of different contents. An error count unit is configured to determine an amount of error between the image and the candidate compressed code for each of the encoders. A mode decision unit is coupled to receive a plurality of the amount of error, the candidate compressed code associated with least amount of error being outputted as an adaptive compressed code for the image.

Abstract: An auto-contrast enhancement system includes a human visual system (HVS)-based local difference (LD) histogram unit configured to build a LD histogram with respect to intensity values; a histogram classifier configured to categorize histograms of input images based on distribution properties; and a histogram equalization (HE) unit configured to process the input image according to a result of the HVS-based LD histogram unit and the enhancement level determined in the histogram classifier.

Abstract: A content adaptive compression system includes a plurality of encoders being coupled to receive a portion of an image, and accordingly generating candidate compressed codes, respectively, the plurality of encoders being configured for encoding images of different contents. An error count unit is configured to determine an amount of error between the image and the candidate compressed code for each of the encoders. A mode decision unit is coupled to receive a plurality of the amount of error, the candidate compressed code associated with least amount of error being outputted as an adaptive compressed code for the image.

Abstract: An auto-contrast enhancement system includes a human visual system (HVS)-based local difference (LD) histogram unit configured to build a LD histogram with respect to intensity values; a histogram classifier configured to categorize histograms of input images based on distribution properties; and a histogram equalization (HE) unit configured to process the input image according to a result of the HVS-based LD histogram unit and the enhancement level determined in the histogram classifier.

Abstract: A gesture recognition system includes a candidate node detection unit coupled to receive an input image in order to generate a candidate node; a posture recognition unit configured to recognize a posture according to the candidate node; a multiple hands tracking unit configured to track multiple hands by pairing between successive input images; and a gesture recognition unit configured to obtain motion accumulation amount according to tracking paths from the multiple hands tracking unit, thereby recognizing a gesture.

Abstract: A system of image stereo matching includes at least one stereo matching unit (SMU) each receives a first view and a second view of a view pair, according to which the SMU generates a first depth map for the first view. The system also includes a backward tracer operable to receive the first depth map, according to which a second depth map for the second view is derived.

Abstract: In a system of object detection, a color detector detects at least one image region in an input image having a color specifically pertinent to the object under detection, thereby obtaining an object width. A dynamic down-sampling unit adaptively performs down-sampling on the detected image region using a generated down-sampling factor according to the object width. An image feature generator receives the down-sampled image and accordingly generates image features for describing the object under detection, and a cascade of classifiers then operates on the image features.