Edge enhancement and smoothing in digital images

• US 5,031,227 A
• Filed: 09/27/1990
• Issued: 07/09/1991
• Est. Priority Date: 09/27/1990
• Status: Expired due to Fees

• Current Assignee: Lockheed Martin Corporation (Martin Marietta Corporation)
• Original Assignee: Loral Aerospace Corp. (MHR Holdings LLC (New York))
• Inventors: Roberts, Gregory A., Raasch, Charles F., Conboy, Mark A.
• Primary Examiner(s): BRINICH, STEPHEN M
• Application Number:
  US07/588,790
• Time in Patent Office: 285 Days
• Field of Search: 382/22, 382/27, 382/54
• US Class Current: 382/199
• CPC Class Codes: G06T 2207/10048 Infrared image
• CPC Class Codes: G06T 2207/20012 Locally adaptive
• CPC Class Codes: G06T 2207/20192 Edge enhancement; Edge preservation
• CPC Class Codes: G06T 5/20 using local operators
• CPC Class Codes: G06T 5/70 Denoising; Smoothing
• CPC Class Codes: G06T 5/73 Deblurring; Sharpening

Abstract

Edge definition in FLIR images is improved by assigning to each pixel (X) the average intensity value of that 3×3 pixel sub-matrix (30a-30i) containing the pixel (X) which exhibits the smallest edge value. Apparatus for obtaining that intensity value in real time is disclosed. This process can be iteratively repeated on the same image, with each iteration creating more enhanced edges and greater area smoothing between edges.

Claims

1. A method of simultaneously smoothing image areas and enhancing their edges in a digitized image, comprising the steps of:

(a) producing successive master pixels of said digitized image;

(b) producing, for each said master pixel, a matrix of pixels centered upon said master pixel;

(c) computing, for each of a plurality of sub-matrices of said matrix each containing said master pixel, the average intensity of said sub-matrix and the edge value thereof, the edge value of a sub-matrix being the maximum absolute difference between the average intensity of that sub-matrix and the average intensities of at least some of the sub-matrices of the same size centered upon a pixel of that sub-matrix;

(d) assigning to said master pixel the average intensity of the sub-matrix within the set of sub-matrices containing said master pixel which has the lowest edge value.

2. The method of claim 1, in which said sub-matrices are all the possible 3×

3 matrices within a 7×

7 matrix.

3. Apparatus for simultaneously smoothing and enhancing the edges of objects in a digitized image, comprising:

(a) image information defining a line-scanned image composed of pixels;

(b) line delay means for simultaneously generating intensity information of a master pixel and pixels adjacent thereto;

(c) average filter means connected to said line delay means for computing the average intensity of a pixel sub-matrix centered upon said master pixel;

(d) average buffer means connected to said average filter means for producing the average intensity values of pixel sub-matrices including said master pixel;

(e) edge image creator means connected to said average buffer means for producing, for each of said sub-matrices, an edge value representing the maximum absolute difference between the average intensity of a given sub-matrix and the average intensities of the sub-matrices centered upon pixels adjacent to the center pixel thereof;

(f) edge buffer means connected to said edge image creator means for producing said edge values for the set of sub-matrices including said master pixel;

(g) neighborhood minimum detector means connected to said edge buffer means for producing an indication of which sub-matrix of said set has the lowest edge value;

(h) latch array means operationally connected to said average filter means for storing the average intensity values of the sub-matrices including said master pixel; and

(i) output means connected to said neighborhood minimum detector means and said latch array means for outputting as the intensity value of said master pixel the average value of the sub-matrix of said set indicated by said neighborhood minimum detector means as having the lowest edge value.

4. The apparatus of claim 3, in which said sub-matrices are 3×

3 pixels in size.