Digital images matching chip

Abstract

The invention discloses a kind of digit image matching circuit structure. The circuit includes address generating module, master slave related process module, comparing localization and control logic generating module, FIFO and exterior controller interface module. The address generating circuit provides exterior memorizing address and reading address for searching area picture elements, and generates interruption instructing signal to exterior controller; the master slave related circuit carries on two-stage buffer to each small module at the same time, and carries on shift registering to image data in searching area; FIFO module memorizes the middle result of matching calculation and adds up; the comparing and control logic module compares and judges the best matching point; the exterior controller interface circuit generates correspondent control signal and outputs each unit circuit respectively. The invention is simple, the matching and searching speed is quick.

Description

technical field [0001] The invention relates to the technical field of basic electronic components, in particular to a chip structure for digital image matching. Background technique [0002] The so-called image matching is a method used to determine the position of a smaller image (usually called a template image) in another larger image (usually called a search area image). It is widely used in pattern recognition, target detection, fire control, sequence image coding, navigation, medical image processing and other systems that require precise positioning of images. [0003] Existing image matching is generally realized by digital signal processor DSP and DSP+FPGA structure. The biggest disadvantage of this matching operation is slow speed, high cost, and it is not conducive to expansion. Foreign Ting-Pang Lin and Chaur-Heh Hsieh once proposed a flexible and fast image matching hardware structure (A Modular and Flexible Architecture for Real-time Image Template Matching, ...

AI Technical Summary

Problems solved by technology

However, when this structure matches the 32×32 template required in practice in the 80×80 search area image, 32 operation processing units PU are needed, and 32 hardware units are gradually used in the early stage of operation, so In the early stage of the operation, many operation processing units PU are performing empty operations, resulting in waste of resources

In addition, the increase in the number of operation processing units PU will inevitably lead to an increase in the resources of the operation control unit PU, and the complexity of control will also increase.

Therefore, this hardware structure has the disadvantages of complex structure and high resource consumption, and can only expand the scope of matching templates and search areas by increasing the scale of hardware.

[0004] In 2002, Zhang Suinan, Huang Shitan and others in China "A High-performance ASIC Design Oriented to Image Matching Algorithm", Journal of Beijing Normal University (Natural Science Edition), February 2002, Volume 38, Issue 48-52 , and proposed an ASIC design for image matching, but this design only completed the hardware implementation of part of the image matching algorithm, while other operations still rely on digital signal processors (DSP) to achieve, and did not complete the entire image matching operation The hardware design of the device, the performance of the image matcher has not been substantially improved

Images