Method of low-pass filtering and arrangement for performing said method

• US 5,237,524 A
• Filed: 06/07/1991
• Issued: 08/17/1993
• Est. Priority Date: 06/29/1990
• Status: Expired due to Fees

• Current Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Original Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Inventors: Heinemann, Hartmut
• Primary Examiner(s): NGUYEN, LONG
• Application Number:
  US07/712,219
• Time in Patent Office: 802 Days
• Field of Search: 364/724.01, 364/724.19, 364/723, 381/46, 358/36, 358/37, 358/54, 358/463
• US Class Current: 708/322
• CPC Class Codes: H03H 21/0012 Digital adaptive filters

Abstract

Signal smoothing can be improved without simultaneously obliterating the useful input signal edges by adaptively filtering the signal, in which the averaging interval is corresponding to the length of a kernel is incremented or decremented as a function of a difference between an output signal sample computed from the kernel and an output signal sample succeeding the kernel.

Claims

1. An image processing method comprising:

scanning, row by row, a two-dimensional picture comprising an array of pixels arranged in columns and rows to obtain a first sequence of signal values, successive signal values of each row of said first sequence corresponding to respective successive pixels of said picture which are arranged in a line direction;

storing said first sequence of signal values into a picture memory;

reading a second sequence of signal values picture memory column by column in a manner that successive signal values of each column correspond to respective successive pixels of said picture which are arranged in a direction perpendicular to the line direction;

generating a third sequence of filtered signal values by deriving each filtered signal value as a function of signal values comprised in a current kernel, said current kernel having a length consisting of L successive picture values of said second sequence and being shifted relative to said second sequence by one signal value position for each filtered signal value derived;

forming a difference between a value derived from a current kernel and a signal value of said second sequence at a position proximate to and not previously comprised in said current kernel;

incrementing the length of the current kernel to no more than a maximum length in case said difference is relatively small; and

decrementing the length of the current kernel to no less than a minimum length in case said difference is relatively large.

2. A method as claimed in claim 1, wherein said difference is formed between a mean value of the signal values comprised in said current kernel and said signal value proximate to the current kernel, said incrementing step is performed when said difference is less than a first threshold value, and said decrementing step is performed when said difference does not fall below a second threshold value which at least as large as the first threshold value.

4. An apparatus as claimed in claim 1, wherein said difference is formed by said processing means between a mean value of the signal values comprised in said current kernel and said signal value proximate to the current kernel, said incrementing by said processing means is when said difference is less than a first threshold value, and said decrementing by said processing means is when said difference does not fall below a second threshold value which at least as large as the first threshold value.

3. An image processing apparatus comprising:

a sensors for scanning, row by row, a two-dimensional picture comprising an array of pixels arranged in rows and columns, to obtain a first sequence of signal values, successive signal values of each row of said first sequence corresponding to respective successive pixels of said picture which are arranged in a line direction;

a picture memory for storing said first sequence of signal and for supplying a second sequence of signal values read from said picture memory column by column in a manner that successive signal values of each column correspond to respective successive pixels of said picture which are arranged in a direction perpendicular to the line direction; and

processing means for generating a third sequence of filtered signal values by deriving each filtered signal value as a function of signal values comprised in a current kernel, said current kernel having a length consisting of L successive picture values of said second sequence and being shifted relative to said second sequence by one signal value position for each filtered signal value derived, for forming a difference between a value derived from a current kernel and a signal value of said second sequence at a position proximate to and not previously comprised in said current kernel, for incrementing the length of the current kernel to no more than a maximum length in case said difference is relatively small, and for decrementing the length of the current kernel to no less than a minimum length in case said difference is relatively large.