Digital image interpolator using a plurality of interpolation kernals

• US 5,125,042 A
• Filed: 05/14/1991
• Issued: 06/23/1992
• Est. Priority Date: 06/16/1989
• Status: Expired due to Term

• Current Assignee: Eastman Kodak Company
• Original Assignee: Eastman Kodak Company
• Inventors: Kerr, Richard, Krogstad, Robert
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): Cammarata, Michael
• Application Number:
  US07/701,627
• Time in Patent Office: 406 Days
• Field of Search: 382/47, 382/41, 358/428, 358/451, 364/723, 364/577, 364/413.18, 364/474.31
• US Class Current: 382/300
• CPC Class Codes: G06F 17/175 of multidimensional data
• CPC Class Codes: G06T 3/10 Selection of transformation methods according to the characteristics of the input images
• CPC Class Codes: G06T 3/4007 based on interpolation, e.g. bilinear interpolation image demosaicing G06T3/4015; edge-driven or edge-based scaling G06T3/403

Abstract

An interpolator for enlarging or reducing a digital image includes an interpolation coefficient memory containing interpolation coefficients representing a one dimension interpolation kernel. A row interpolator receives image pixel values, retrieves interpolation coefficients from the memory, and produces interpolated pixel values by interpolating in a row direction. A column interpolator receives multiple rows of interpolated pixel values from the row interpolator, retrieves interpolation coefficients from the memory, and produces rows of interpolated pixel values by interpolating in a column direction.

Claims

1. Digital image processing apparatus for performing two dimensional interpolation on a digital image expressed as rows and columns of pixel values, comprising:

(a) an interpolation coefficient memory containing a plurality of one dimensional interpolation kernels for performing a plurality of interpolation types including cubic convolution, linear, and replication interpolation;

(b) row interpolator means connected to said memory for receiving successive image pixel values along a row, retrieving interpolation coefficients from said memory, and producing interpolated pixel values by multiplying and summing the pixel values and interpolation coefficients;

(c) column interpolator means connected to said row interpolator means and said memory for receiving rows of interpolated and noninterpolated pixel values from said row interpolator means, retrieving interpolation coefficients from said memory, and producing rows of interpolated pixel values by interpolating between groups of interpolated and noninterpolated pixel values in a column direction;

(d) logic and control means for selecting one of said kernels in response to an operator input; and

wherein said row interpolator means includes means for replicating the first and last pixel value in a row in an image, and said column interpolator means includes means for replicating the first and last rows of interpolated and noninterpolated pixel values in an image.

5. the apparatus claimed in claim 1, wherein said interpolation coefficient memory is divided into a plurality of banks adapted to be addressed simultaneously to return a plurality of interpolation coefficients, one such coefficient from each of said banks.

6. The apparatus claimed in claim 1, wherein said row interpolator means further comprises row controller means for requesting pixel values from a frame store memory and interpolation coefficient values from the coefficient memory, and row arithmetic and logic unit (ALU) means for receiving the pixel values and the coefficient values, and performing multiplication summation according to the formula P'"'"'=(P₁ ×

C₁)+(P₂ ×

C₂)+(P₃ ×

C₃)+(P₄ ×

C₄) where P'"'"' is an interpolated value, P_i are the pixel values and C_i are the coefficient values.

7. The apparatus claimed in claim 6, wherein said row ALU means comprises a four element input register for receiving said pixel values serially, a four element working register coupled to said input register for receiving said pixel values in parallel from said input register, a four element coefficient register for receiving interpolation coefficients in parallel from said coefficient memory, and multiply accumulation means coupled to said working register and said coefficient register for receiving said pixel values and coefficient values in pairs, multiplying them, and accumulating the sum of their products.

8. The apparatus claimed in claim 1, wherein said column interpolator means further comprises line store means for receiving rows of interpolated and noninterpolated pixel values from said row interpolator, line store control means for controlling the addressing of said line stores, ALU means for retrieving interpolated and noninterpolated pixel values from said line stores, and interpolating between the rows in a column direction, and column controller means for controlling the column interpolator means.

9. The apparatus claimed in claim 8, wherein said column ALU means comprises a 4 element coefficient register for receiving, in parallel, four interpolation coefficients from the coefficient memory and delivering them serially, and column multiply and accumulation means connected to said coefficient register and having inputs for receiving successive pixel values and coefficient values, and for performing a multiplication and summing the products to produce an interpolated pixel value according to the equation P"=(P'"'"'₁,1 ·

C₁)+(P'"'"'₁,2 ·

C₂)+(P'"'"'₁,3 ·

C₃)+(P'"'"'₁,4 ·

C₄) where P" is the interpolated pixel value between rows of pixels, P'"'"'_ij is the i^th pixel value from the j^th row of interpolated values, and C_k are the interpolation coefficients.

10. The apparatus claimed in claim 8, wherein said line store means comprises five first-in first-out line memories, multiplexing means for coupling the outputs of said memories in groups of 4 to the ALU means, and read/write select means for controlling said multiplexing means and for controlling the data written into said line store means.

2. A method of processing a digital image comprising the steps of:

selecting a format of an output digital image comprising at least first and second image bands, each of said bands having at least one digital image;

determining first and second different interpolation parameters for said respective first and second image bands;

processing a first digital image to be incorporated in said first image band of an output digital image with said first interpolation parameters;

processing a second digital image to be incorporated in said second image band of said output digital image with said second interpolation parameters; and

wherein said steps of processing said first and second digital images are effected by means of two-dimensional interpolation including;

storing a plurality of one dimensional interpolation kernels for performing a plurality of interpolation types including cubic convolution, linear, and replication interpolation in an interpolation coefficient memory;

performing row interpolation by receiving successive image pixel values along a row, retrieving interpolation coefficients from said memory, and producing interpolated pixel values by multiplying and summing the pixel values and interpolation coefficients;

performing column interpolation by retrieving interpolation coefficients from said memory, and producing rows of interpolated pixel values by interpolating between groups of interpolated and noninterpolated pixel values in a column direction; and

controlling the selection of one of said kernels in response to an operator input.

3. The method of claim 2 wherein said determining steps includes determining first and second different horizontal and vertical interpolation sampling increments for said respective first and second image bands and wherein said respective steps of processing said first and second digital images are effected with said determined first and second different horizontal and vertical interpolation and sampling increments.

4. The method of claim 2 wherein said selecting step includes selecting an output digital image formed which further includes a text band having text and including the step of processing said text to be incorporated in said text band by means of replication interpolation.