摘要:
A method of color image processing for quantizing output includes obtaining an input for an object pixel which is represented by a vector in a first color space. A modified input equal to the input plus a sum of errors from other pixels in a neighborhood of the object pixel is generated. For each color component in the first color space, where corresponding color components of the modified input are located with respect to a preset range is determined. If the modified input's color component is greater than the preset range, then that color component for an output is determined to be on; if less than the preset range, then that color component for the output is determine to be off; and, if within the preset range, then that color component for the output is determined to be unknown. A transformed modified input is mapped to a perceptual color space when any color component of the output is unknown. Colors consistent with color components of the output that have already been determined are also mapped to the perceptual color space. The color in the perceptual color space that lies closest to the transformed modified input is chosen. An output in the first color space having color components on and off is generated consistent with the determinations and/or choices made. Error for the object pixel is then calculated as the difference between the output and the modified input.
摘要:
A method, a device and a computer readable for automatically identifying a Christmas tree scene and setting a camera's focus and/or exposure parameters in a way that yields images with high image quality. The Christmas tree scene identification can be performed by segmenting the image into bright and dark regions, identifying the light objects, collecting the statistics of the light objects, and classifying the scene based on the statistics of the light objects, or by collecting the pixel value statistics for the image, and classifying the scene based on the statistics of the pixel values, or by collecting the pixel value statistics for the image, filtering the image, collecting the pixel value statistics for the filtered image, and classifying the scene by comparing the statistics of the pixel values before and after filtering. The focus and exposure settings can be adjusted based on the Christmas tree scene identification results. For Christmas tree scenes, the exposure can be set based on a value that is adjusted upwards from the mean luminance of the image, or on a value that is calculated from the top luminance value. The focus can be set by identifying the lights in the image, and minimizing the light size in the image.
摘要:
A method, a mobile image capturing device and a computer readable for capturing and processing both document and non-document images in optimized manners. The present invention contains steps: a) determining if an image to be captured is a document image or a non-document image; b) capturing and processing said image with methods and parameters optimized for document images if said determination is document; c) capturing and processing said image with methods and parameters optimized for non-document images if said determination is non-document.
摘要:
A method and a system for automating parking payment includes receiving video data in a sequence of frames provided by a video capture device observing a parking area. The method includes detecting a vehicle in the parking area using the video data. The method further includes receiving information of a vehicle in the parking area provided by a user device. The method includes comparing the characteristics of the detected vehicle with the information from the user device. In response to the characteristics of the detected vehicle being a match to the information sent from the user device, the method includes tracking the detected vehicle across the sequence of frames. The method includes computing a duration that the detected vehicle remains stationary using the tracking. The method includes computing an amount for charging the associated user device based on the duration.
摘要:
A method and apparatus are disclosed for embedding a two-dimensional machine-readable code into the background of a document in an aesthetically pleasing manner. The objects (picture, graphics, text, etc.) in a document image are first segmented from the background. A location for embedding the barcode is determined. A texture image is created by replicating the barcode to a size that covers the entire image. A modulation index map is generated from the segmentation results. The original image is modulated to produce the output image based on the texture image and the modulation index map.
摘要:
A method for determining a parking violation includes receiving video data as a sequence of frames provided by a camera. The method includes defining a location of an exclusion zone in the video data. The method includes detecting a vehicle located in the defined exclusion zone. The detecting includes determining a background in an initial frame of the video data and determining a background in a select frame by applying a predetermined updating process. The detecting includes subtracting the background of the select frame from the initial frame to obtain an image difference. The detecting includes classifying the pixels in the image difference as foreground or background pixels and classifying the pixels in the foreground image as vehicle or non-vehicle pixels. The method includes determining a duration that the detected vehicle is in the exclusion zone based on a number of the sequence of frames including the detected vehicle.
摘要:
A method for adjusting a license plate that is detected in a captured image includes automatically determining at least one set of correction parameters corresponding to a slant-oriented license plate. The method further includes receiving an input image representing a detected license plate. In response to receiving the input image, the method includes automatically adjusting the input image to obtain a corrected image using the at least one set of correction parameters.
摘要:
What is disclosed is a novel system and method to control an imaging device. A first collection of a plurality of two-dimensional image points in a first vector space for a source image and a second collection of a plurality of two-dimensional image points in a second vector space for a target image are received. The first collection of two-dimensional image points and the second collection of two-dimensional image points are converted into a homogenous form to apply affine matrix transformations. A matrix transformation is solved to map the first collection of the plurality of two-dimensional image points to the second collection of the plurality of two-dimensional image points. The matrix transformation is used to adjust distortion of the target image in an imaging device.
摘要:
A method and a system for identifying a vehicle in a parking area includes receiving video data from a video camera observing a parking area from a first field of view over duration of time and receiving image data from a still camera observing a second field of view overlapping the first field of view. The method includes tracking a location of a vehicle across a sequence of frames. The tracking includes determining a time instant at which each frame was captured. The method includes comparing a select frame captured by the video camera with an image captured by the still camera. Based on the comparing, the method includes matching the vehicle in the image with the vehicle in the frame. The method includes determining a license plate number of the vehicle by locating a license plate on the vehicle in the image.