Abstract: An electrocardiography signal extraction method includes receiving an electrocardiography signal, performing a time-frequency transformation on the received electrocardiography signal to generate a corresponding scalogram, selecting a pre-defined R-pertinent scale, performing the time-frequency transformation at the selected pre-defined R-pertinent scale to generate a R-pertinent summarized response, obtaining a R peak position, selecting a pre-defined QRS-pertinent scale, performing the time-frequency transformation at the selected pre-defined QRS-pertinent scale, obtaining a Q peak position and a S peak position of the electrocardiography signal by finding relative maximum negative responses before and behind the R peak position respectively, obtaining a QRSon position and a QRSoff position by finding relative minimum second derivatives of the responses before the Q peak position and behind the S peak position, respectively.

Abstract: An electrocardiography signal extraction method includes receiving an electrocardiography signal, detecting a peak of a wave of the electrocardiography signal, separating the wave into left and right waves, normalizing the left wave and a plurality of scales of Gaussian, comparing the normalized left wave with a left part of the normalized scales of Gaussian, acquiring a left part error function, indicating a left minimum comparative error, selecting a left scale of Gaussian with the left minimum comparative error, obtaining a left duration of the wave, normalizing the right wave, comparing the normalized right wave with a right part of the normalized scales of Gaussian, acquiring a right part error function, indicating a right minimum comparative error, selecting a right scale of Gaussian with the right minimum comparative error, obtaining a right duration of the wave, and obtaining an extracted wave.

Abstract: A design space exploration method of a reconfigurable motion compensation architecture is disclosed.

Abstract: This invention discloses a method for determining an area scaling ratio of an object in a video image sequence. In one embodiment, a centroid of the object is determined. One or more directed straight lines are selected, each passing through the centroid, extending from an end of the object's boundary to an opposite end thereof, and having a direction that is upward. A length scaling ratio for each directed straight line is determined by: determining a motion vector for each selected pixel on the line; computing a scalar component of the motion vector projected onto the line; estimating a change of the line's length according to the scalar components obtained for all pixels; and determining the length scaling ratio according to the change of the line's length. The area scaling ratio is computed based on the length scaling ratios for all directed straight lines.

Abstract: A method of determining a design framework is implemented by an algorithm analyzer. The method includes configuring the algorithm analyzer to perform intrinsic complexity analysis of an algorithm for a predetermined application to obtain a set of parameters representing intrinsic characteristics of the algorithm. The method also includes configuring the algorithm analyzer to establish candidate design frameworks based on the parameters. Each candidate design framework includes a set of design constraints corresponding to the algorithm and which are used when designing a hardware and/or software configuration for implementing the predetermined application. The method also includes configuring the algorithm analyzer to analyze the suitability of the set of design constraints of each candidate design framework based on given specification restrictions of the predetermined application to determine which candidate design framework(s) is suited for the predetermined application.

Abstract: A method of analyzing intrinsic parallelism of an algorithm, comprising: generating a dataflow graph which is composed of vertexes representing computation and directed edges denoting the dependency and flow of data from the algorithm; building a matrix representing the dataflow graph; and quantifying the intrinsic parallelism based on rank and dimension of the matrix representing the generated dataflow graph.

Abstract: Methods for decoding are provided. The proposed method includes steps of: receiving a most probable symbol (MPS) value and a probability value for generating a probability model update; and receiving the probability model update for generating the MPS value and the probability value, wherein when the probability value shows that an MPS is occurred, a path corresponds to the MPS is estimated and a first bin included in the path is decoded beforehand.

Abstract: The quantifying method for intrinsic data transfer rate of algorithms is provided. The provided quantifying method for an intrinsic data transfer rate includes steps of: detecting whether or not a datum is used; providing a dataflow graph G including n vertices and m edges, and a Laplacian matrix L having ixj elements L(i,j) when the datum is not reused, wherein each of the vertices represents one of an operation and a datum, each of the edges represents a data transfer, and vi is the ith vertex; and using the Laplacian matrix L to estimate a maximum quantity of the intrinsic data transfer rate.

Abstract: A motion estimation method includes: (A) defining one pixel in a reference image as a center of search (CS) corresponding to a target pixel set in a current image; (B) determining a center error (CE) signal; (C) defining another pixel in the reference image as a target of search (TS) with reference to the CS, one candidate search vector available for selection from a vector set, and a step size; (D) determining a target error (TE) signal; (E) determining whether to update the CS and the CE signal; (F) if determined, updating the CS, the CE signal and the vector set; (G) repeating steps (C)˜(F) using a candidate search vector selected from the vector set and the same step size until there is no candidate search vector available for selection in the vector set; (H) repeating steps (C)˜(G) using a smaller step size until a predetermined value is reached; and (I) computing a motion vector based on the target pixel set and one pixel set that includes the CS.

Abstract: Methods for decoding are provided. The proposed method includes steps of: receiving a most probable symbol (MPS) value and a probability value for generating a probability model update; and receiving the probability model update for generating the MPS value and the probability value, wherein when the probability value shows that an MPS is occurred, a path corresponds to the MPS is estimated and a first bin included in the path is decoded beforehand.

Abstract: A design space exploration method of a reconfigurable motion compensation architecture is disclosed.

Abstract: A method and system of extracting a perceptual feature set for image/video segmentation are disclosed. An input image is converted to obtain a hue component and a saturation component, where the hue component is quantized into a number of quantum values. After weighting the quantized hue component with the saturation component, the weighted quantized hue component and the saturation component are subjected to a statistical operation in order to extract feature vectors. Accordingly, the method and system provide overall segmentation results that are very close to human interpretation.

Abstract: Memory storage requirements for digital signal processing operations, for example, motion-compensated video scan rate conversion, that produce intermediate output data, which is then used as an input to the operation, are reduced by reordering operations and organizing memory allocations in a special manner to allow intermediate output at a particular execution time, to substantially share the same memory space as the intermediate output of a previous execution time. Such a reduction in the amount of memory required for processing operations advantageously reduces cost and power consumption.

Abstract: A method of analyzing intrinsic parallelism of an algorithm, comprising: generating a dataflow graph which is composed of vertexes representing computation and directed edges denoting the dependency and flow of data from the algorithm; building a matrix representing the dataflow graph; and quantifying the intrinsic parallelism based on rank and dimension of the matrix representing the generated dataflow graph.

Abstract: A method of determining a design framework is implemented by an algorithm analyzer. The method includes configuring the algorithm analyzer to perform intrinsic complexity analysis of an algorithm for a predetermined application to obtain a set of parameters representing intrinsic characteristics of the algorithm. The method also includes configuring the algorithm analyzer to establish candidate design frameworks based on the parameters. Each candidate design framework includes a set of design constraints corresponding to the algorithm and which are used when designing a hardware and/or software configuration for implementing the predetermined application. The method also includes configuring the algorithm analyzer to analyze the suitability of the set of design constraints of each candidate design framework based on given specification restrictions of the predetermined application to determine which candidate design framework(s) is suited for the predetermined application.

Abstract: Memory storage requirements for digital signal processing operations, for example, motion-compensated video scan rate conversion, that produce intermediate output data, which is then used as an input to the operation, are reduced by reordering operations and organizing memory allocations in a special manner to allow intermediate output at a particular execution time, to substantially share the same memory space as the intermediate output of a previous execution time. Such a reduction in the amount of memory required for processing operations advantageously reduces cost and power consumption.

Abstract: A motion estimation method includes: (A) defining one pixel in a reference image as a center of search (CS) corresponding to a target pixel set in a current image; (B) determining a center error (CE) signal; (C) defining another pixel in the reference image as a target of search (TS) with reference to the CS, one candidate search vector available for selection from a vector set, and a step size; (D) determining a target error (TE) signal; (E) determining whether to update the CS and the CE signal; (F) if determined, updating the CS, the CE signal and the vector set; (G) repeating steps (C)˜(F) using a candidate search vector selected from the vector set and the same step size until there is no candidate search vector available for selection in the vector set; (H) repeating steps (C)˜(G) using a smaller step size until a predetermined value is reached; and (I) computing a motion vector based on the target pixel set and one pixel set that includes the CS.

Abstract: A method and system of mono-view depth estimation are disclosed. A two-dimensional (2D) image is first segmented into a number of objects. A depth diffusion region (DDR), such as the ground or a floor, is then detected among the objects. The DDR generally includes a horizontal plane. The DDR is assigned the depth, and each object connected to the DDR is assigned depth according to the depth of the DDR at the connected site.

Abstract: A method and system of extracting a perceptual feature set for image/video segmentation are disclosed. An input image is converted to obtain a hue component and a saturation component, where the hue component is quantized into a number of quantum values. After weighting the quantized hue component with the saturation component, the weighted quantized hue component and the saturation component are subjected to a statistical operation in order to extract feature vectors. Accordingly, the method and system provide overall segmentation results that are very close to human interpretation.

Abstract: According to an example embodiment, the present invention is directed to pixel-data processing that includes scanning a first 2×2 line in each of a series of immediately adjacent pixel blocks, prior to scanning a second 2×2 line in each of the series of pixel blocks. Each scanned line is then processed for motion compensation in a manner that addresses challenges, including those discussed above, related to buffer size requirements, power consumption requirements and latency.