Abstract: A particle filter is suitable for performing particle filtering on a frame to track a particular object in the frame. The particle filter includes a frame cache, an observation model generator, and a particle filter controller. The frame cache is connected to a system memory through a system bus, in which the system memory stores all image blocks of the frame; and the frame cache obtains the at least one image block of the frame from the system memory and stores the obtained image block. The observation model generator reads at least one pixel from the frame cache, and generates an observation model corresponding to the object and the read image block according to the read pixel. The particle filter controller obtains the observation model from the observation model generator, and determines and outputs an object tracking result of the object according to the observation model.

Abstract: A particle filter is suitable for performing particle filtering on a frame to track a particular object in the frame. The particle filter includes a frame cache, an observation model generator, and a particle filter controller. The frame cache is connected to a system memory through a system bus, in which the system memory stores all image blocks of the frame; and the frame cache obtains the at least one image block of the frame from the system memory and stores the obtained image block. The observation model generator reads at least one pixel from the frame cache, and generates an observation model corresponding to the object and the read image block according to the read pixel. The particle filter controller obtains the observation model from the observation model generator, and determines and outputs an object tracking result of the object according to the observation model.

Abstract: A digital video target moving object segmentation method and system is designed for processing a digital video stream for segmentation of every target moving object that appears in the video content. The proposed method and system is characterized by the operations of a multiple background imagery extraction process and a background imagery updating process for extracting characteristic background imagery whose content includes the motional background objects in addition to the static background scenes; and wherein the multiple background imagery extraction process is based on a background difference threshold comparison method, while the background imagery updating process is based on a background-matching and weight-counting method. This feature allows an object mask to be defined based on the characteristic background imagery, which can mask both the motional background objects as well as the static background scenes.

Abstract: A module for generating real-time, multiple-resolution video streams and the architecture thereof are disclosed. A module for generating multiple-resolution video streams as well as the architecture thereof for use with a video encoder includes a system bus, an external memory and a main processor. The main processor and the external memory are coupled to the system bus. The main processor includes a microprocessor, a main arithmetic unit and a secondary arithmetic unit. By applying the present invention, a less time-consuming arithmetic module can synchronously perform together with a more time-consuming arithmetic module, thereby reducing idle time and increasing hardware efficiency and parallelism.

Abstract: A reconfigurable image processor for image processing includes an arithmetic module, a first memory unit, a bus control module and a connecting module. By setting different configurations or configuring connections among arithmetic units via the connecting module, the operations of arithmetic units are reconfigured to include different functions. The present invention provides an application architecture including a sensor module, a display module, a second memory unit and a reconfigurable image processor.

Abstract: A digital video target moving object segmentation method and system is designed for processing a digital video stream for segmentation of every target moving object that appears in the video content. The proposed method and system is characterized by the operations of a multiple background imagery extraction process and a background imagery updating process for extracting characteristic background imagery whose content includes the motional background objects in addition to the static background scenes; and wherein the multiple background imagery extraction process is based on a background difference threshold comparison method, while the background imagery updating process is based on a background-matching and weight-counting method. This feature allows an object mask to be defined based on the characteristic background imagery, which can mask both the motional background objects as well as the static background scenes.

Abstract: A module for generating real-time, multiple-resolution video streams and the architecture thereof are disclosed. A module for generating multiple-resolution video streams as well as the architecture thereof for use with a video encoder includes a system bus, an external memory and a main processor. The main processor and the external memory are coupled to the system bus. The main processor includes a microprocessor, a main arithmetic unit and a secondary arithmetic unit. By applying the present invention, a less time-consuming arithmetic module can synchronously perform together with a more time-consuming arithmetic module, thereby reducing idle time and increasing hardware efficiency and parallelism.

Abstract: A reconfigurable image processor for image processing includes an arithmetic module, a first memory unit, a bus control module and a connecting module. By setting different configurations or configuring connections among arithmetic units via the connecting module, the operations of arithmetic units are reconfigured to include different functions. The present invention provides an application architecture including a sensor module, a display module, a second memory unit and a reconfigurable image processor.

Abstract: Provided is an external memory accessing architecture for use with IC comprising a first bus connected to an external memory and having n-bit data width; a first buffer unit of k serially connected first buffers each having n-bit data width, a first one of the first buffers connected to the external memory via the first bus; a second buffer unit comprising a second buffer having k*n-bit data width, the second buffer connected to the first buffers; a second bus connected to the second buffer for transferring k*n-bit data; an output unit connected to the second buffer unit and comprising a multiplexer; and a controller connected to the output unit, the second bus, and the external memory respectively such that the controller is able to read data from the external memory or transfer data thereto via the second bus and at least one control signal in higher transfer rate.

Abstract: A method for motion estimation for use in video coding. In a predetermined search region, three starting rows are chosen, wherein a given prediction point is on the middle row, the N-th row. The method includes the following steps. (a) All points of the (N−2)-th, N-th, and (N+2)-th rows are chosen as checking points. (b) Block match errors for the checking points are calculated and a checking point which corresponds to a minimum block match error is defined as a candidate checking point. (c) A determination is made whether the candidate checking point is on the N-th row; if so, the method proceeds to step (g); otherwise, it proceeds to step (d). (d) It is determined that whether the candidate checking point is on the (N+2)-th row; if so, the method proceeds to step (e); otherwise, it proceeds to step (f). (e) N is incremented by two, all points of the (N+2)-th row are chosen as checking points, and step (b) is repeated.

Abstract: A method for motion estimation for use in video coding. In a predetermined search region, three starting rows are chosen, wherein a given prediction point is on the middle row, the N-th row. The method includes the following steps. (a) All points of the (N−2)-th, N-th, and (N+2)-th rows are chosen as checking points. (b) Block match errors for the checking points are calculated and a checking point which corresponds to a minimum block match error is defined as a candidate checking point. (c) A determination is made whether the candidate checking point is on the N-th row; if so, the method proceeds to step (g); otherwise, it proceeds to step (d). (d) It is determined that whether the candidate checking point is on the (N+2)-th row; if so, the method proceeds to step (e); otherwise, it proceeds to step (f). (e) N is incremented by two, all points of the (N+2)-th row are chosen as checking points, and step (b) is repeated.