Changes To FLCC
Not logged in
     2      2   <font size="5">
     3      3   <b>FLCC:</b>
     4      4   <i>Fast computation of Local Correlation Coefficients</i>
     5      5   </font>
     6      6   </p>
     7      7   
     8      8   
     9         -<h3>Why Use FLCC?</h3>
    10         -
    11      9   <table border="0" cellpadding="1" bgcolor="#558195" align="right" hspace="10">
    12     10   <tr><td>
    13     11   <table border="0" cellpadding="10" bgcolor="white">
    14     12   <tr><td>
    15     13   <h3>Quick Links</h3>
    16     14   <ul>
    17     15   <li> [http:doc/tip/doc/FLCC_Manual_v1.0.pdf | Documentation (pdf)]
................................................................................
    22     20   </td></tr>
    23     21   <tr><td>
    24     22   <center><img src="fossil3.gif"></center>
    25     23   </td></tr>
    26     24   </table>
    27     25   </table>
    28     26   
    29         -<p>
           27  +
    30     28   The FLCC Library (Fast LCC) is a programming tool for fast computation
    31     29   of Local Correlation Coefficients (LCCs) between an image (or a stream
    32     30   of images) and an image template. The LCC computation is a basic
    33     31   image-based information processing step which finds numerous
    34     32   applications in areas such as template or pattern matching, image
    35     33   registration, motion detection and many more. However, the LCC
    36     34   computation has always been considered of high arithmetic complexity and
    37     35   time consuming, especially for real-time applications, thus making the
    38     36   usage of LCCs rather troublesome.
    39     37   
    40         -<p>
           38  +
    41     39   This library intends to overcome this problem and provide users with a
    42     40   simple yet powerful interface for computing the LCCs. Most
    43     41   implementations so far tried to reduce computation time by sacrificing
    44     42   the local normalization characteristic of the LCCs or approximating the
    45     43   result in other lossy ways. This library, though, manages to reduce
    46     44   computation time to a minimum, yet without making any compromises on the
    47     45   quality of the result. The user can be sure that the output result is
    48     46   accurately the “real” LCC distribution in any case, as exactly it is
    49     47   defined formally.
    50     48   
    51         -<p>
           49  +
    52     50   The performance benefit is achieved by two routes. On the one hand, the
    53     51   library implements a set of optimized fast algorithms, selecting the
    54     52   most appropriate for each case. On the other hand, it fully exploits
    55     53   current top-notch architectures, namely multicore CPU processors and
    56     54   manycore GPU devices. In other words, the library takes advantage of the
    57     55   system computational resources, executing in parallel on multiple CPU
    58     56   threads or having the work load carried out by the powerful GPU devices
    59     57   of the system (according to what it deems to be faster).
           58  +
           59  +--
    60     60