초록 ▼

A demosaic system and method that supports multiple CFA pattern inputs is disclosed. The demosaic system is capable of handling both RGB Bayer input and CMYG input and perform demosaic operations on both inputs to recover full-color images from the raw input images. The system uses a variable number

A demosaic system and method that supports multiple CFA pattern inputs is disclosed. The demosaic system is capable of handling both RGB Bayer input and CMYG input and perform demosaic operations on both inputs to recover full-color images from the raw input images. The system uses a variable number gradient demosiac process. The process uses a 5��5 neighborhood of sensor pixel data centered at the pixel under consideration. The process calculates a set of gradients corresponding to different directions within the neighborhood of the sensor pixel data. A threshold value is determined and a subset of gradients is selected from the set of gradients that fall below the threshold value. The system calculates estimation values for the missing color value and the actual measured center pixel color value obtained from the sensor data on directions that are within the subset of gradients below the threshold. The system then determines the sum of the missing color estimation values and the sum of the actual center pixel color estimation values. The system interpolates the missing color value by using the average difference of the summed estimation values for the missing color and the summed estimation values for the actual center measured pixel value from the sensor data.

대표청구항 ▼

The invention claimed is: 1. A demosaic system for converting a sensor image into a full-color image comprising: a gradient module for calculating a plurality of gradients from a neighborhood of the sensor image data centered at a pixel under consideration, wherein a gradient corresponds to a direc

The invention claimed is: 1. A demosaic system for converting a sensor image into a full-color image comprising: a gradient module for calculating a plurality of gradients from a neighborhood of the sensor image data centered at a pixel under consideration, wherein a gradient corresponds to a direction; a threshold module for determining a threshold value and selecting a subset of gradients from the plurality of gradients below the threshold value; an estimation module for calculating an estimation value from the sensor image data for a missing color and an estimation value for a center pixel on a direction in the plurality of gradients and for calculating a sum of a missing color estimation value and a sum of a center pixel estimation value; and a color interpolation module for interpolating a missing color value using an average difference between the sum of the missing color estimation value and the sum of the center pixel estimation value. 2. The demosaic system of claim 1, wherein the neighborhood of the sensor image data comprises a 5��5 array of adjacent pixels centered at the pixel under consideration. 3. The demosaic system of claim 1, wherein the threshold module determines the threshold according to the equation ki*Gmin(i, j)+k2*(Gmax(i, j)-Gmin(i, j)), wherein k1 and k2 are constants and Gram(i, j) is the minimum gradient value in the set of gradients and Gmax(i, j) is the maximum gradient value in the plurality of gradients. 4. The demosaic system of claim 3, wherein k1 is greater than 1 and k2 is less than 1. 5. The demosaic system of claim 4, wherein k1=1.5 and k2=0.5. 6. The demosaic system of claim 1, wherein the sensor pixel data is mapped to four separate colors in alternating patterns in a matrix representation centered at the pixel under consideration. 7. The demosaic system of claim 6, wherein the matrix representation is a 5��5 matrix centered at the pixel under consideration. 8. The demosaic system of claim 1, wherein the plurality gradients is determine from eight directions comprising of the north, northwest, northeast, south, southwest, southeast, west and east directions. 9. The demosaic system of claim 1, wherein the missing color value is calculated by adding an actual measured center pixel color value to the average difference of the sum of the missing color estimation values and the sum of the center pixel estimation values. 10. The demosaic system of claim 1, wherein the average difference is determined by dividing the difference of the sum of the missing color estimation values and the sum of the center pixel estimation values by the number of elements in the subset of gradients. 11. The demosaic system of claim 1, wherein the gradient for each direction is determine determined by summing the absolute value of differences between pairs of similar-colored pixels. 12. The demosaic system of claim 1, wherein the sensor image data comprises an RGB Bayer pattern. 13. The demosaic system of claim 1, wherein the sensor image data comprises a CMYG pattern. 14. A method for performing a demosaic operation on a sensor image data comprising: calculating a plurality of gradients from a neighborhood of the sensor image data centered at a pixel under consideration, wherein a gradient corresponds to a direction; determining a threshold value; selecting a subset of gradients from the plurality of gradients below the threshold value; calculating an estimation value from the sensor image data for a missing color and an estimation value for a center pixel on a direction in the plurality of gradients; calculating a sum of a missing color estimation value and a sum of a center pixel estimation value; and interpolating a missing color value using an average difference between the sum of the missing color estimation value and the sum of the center pixel estimation value. 15. The method of claim 14, wherein an RGB sensor image data is inherently supported. 16. The method of claim 14, wherein a CMYG sensor image data is inherently supported. 17. A method for performing a demosaic operation on a sensor image data comprising: representing the sensor image data using a plurality of color data in alternating patterns in a matrix representation centered at a pixel under consideration; calculating as a plurality of gradients for the plurality of color data centered at the pixel under consideration, wherein a gradient corresponds to a direction; determining a threshold value; selecting a subset of gradients from the plurality of gradients below the threshold value; calculating an estimation value from the plurality of color data for a missing color and an estimation value for a center pixel on a direction in the subset of gradients; calculating a sum of a missing color estimation value and a sum of a center pixel estimation value; and interpolating a missing color value using an average difference between the sum of the missing color estimation value and the sum of the center pixel estimation value. 18. The demosaic method of claim 17, wherein the plurality of colors comprises four separate colors. 19. The demosaic method of claim 17, wherein the sensor image data is an RGB Bayer pattern. 20. A method for processing sensor image data to recover full color image data, comprising: retrieving the sensor image data stored in a memory; using a gradient function, calculating a plurality of gradients from a neighborhood of the sensor image data centered at a pixel under consideration, wherein a gradient corresponds to a direction; using a threshold function, determining a threshold value and selecting a subset of gradients from the plurality of gradients below the threshold value; using an estimation function, calculating an estimation value from the sensor image data for a missing color and an estimation value for a center pixel on a direction in the subset of gradients and calculating a sum of a missing color estimation value and a sum of a center pixel estimation value; and using a color interpolation function, interpolating a missing color value using an average difference between the sum of the missing color estimation value and the sum of the center pixel estimation value. 21. The method of claim 20, wherein the sensor image data is an RGB Bayer pattern. 22. The method of claim 20, wherein the sensor image data is a CMYG pattern.