Auto-focus image system

Abstract

An auto-focus image system includes a pixel array and a focus signal generator coupled to the pixel array. The pixel array captures an image that has more than one edge. The generator generates a focus signal. Each edge has a width and contributes a quantity to compute the focus signal. The quantity may remain unchanged if all pixel values that enter a computation of the quantity are scaled up by a common multiplier. The quantity may have a dimension based solely on a dimension of length. The quantity may be the width. The generator eliminates an edge whose gradient profile has a first width at a first percentage height of the gradient profile that falls outside a tolerance region. The tolerance region varies with a second width of the gradient profile at a second percentage height. Alternatively, the edge's contribution to the focus signal is deemphasized.

1 . A method to generate a focus signal from a plurality of edges in an image to indicate a degree of image sharpness in the image, each edge contributing a quantity, comprising:
 rejecting or attenuating a contribution of said quantity of an edge towards generating said focus signal where a first parameter of a gradient profile of said edge measured at a first gradient level does not meet a predetermined criterion that varies with a second parameter of said gradient profile of said edge measured at a second gradient level,   wherein said first and second gradient levels are different gradient values defined as a first fraction and a second fraction, respectively, of a peak gradient value of said edge, and both said first and second parameters are determined by said gradient profile to both sides of a peak of said gradient profile.   
     
     
         2 . The method of  claim 1 , further comprising:
 evaluating the first parameter and the second parameter to obtain a first value and a second value, respectively; and   determining whether the first value meets the predetermined criterion given the second value.   
     
     
         3 . The method of  claim 1 , wherein the predetermined criterion requires the first parameter to be, to within a predetermined tolerance, the second parameter multiplied by a predetermined multiplier. 
     
     
         4 . The method of  claim 3 , wherein the predetermined multiplier is a function of an edge width of the edge. 
     
     
         5 . The method of  claim 3 , wherein the predetermined multiplier is interpolated from a sequence of multipliers, each multiplier for a different edge width. 
     
     
         6 . The method of  claim 1 , wherein the first parameter is either:
 a width of the gradient profile measured at the first gradient level; or,   a count of pixels in the gradient profile that have gradient levels anywhere from the first gradient level to a peak gradient level of the gradient profile.   
     
     
         7 . The method of  claim 6 , wherein the second parameter is either:
 a width of the gradient profile measured at the second gradient level; or,   a count of pixels in the gradient profile that have gradient levels anywhere from the second gradient level to the peak gradient level.   
     
     
         8 . The method of  claim 6 , wherein the width is measured from a gradient or interpolated gradient on one side of the peak to another gradient or interpolated gradient on the other side of the peak. 
     
     
         9 . The method of  claim 1 , wherein the first and second gradient levels are within a range from 15% to 85% of the peak gradient level of the gradient profile. 
     
     
         10 . The method of  claim 1 , wherein the first and second gradient levels are within a range from 20% to 80% of the peak gradient level of the gradient profile. 
     
     
         11 . The method of  claim 1 , wherein the gradient profile is a sequence of consecutive all-positive or all-negative gradients across a plurality of pixels arrayed side-by-side in a first direction plotted against a sequence of consecutive integers, a peak gradient within the sequence has a peak gradient value whose magnitude is maximal among magnitudes of gradient values of all gradients within the sequence, and each side of the peak gradient includes at least a gradient. 
     
     
         12 . The method of  claim 3 , wherein the edge width is corrected for a slant of the edge for determining the predetermined criterion. 
     
     
         13 . The method of  claim 1 , wherein the predetermined criterion depends on a slant of the edge. 
     
     
         14 . A method to generate a focus signal from a plurality of edges in an image to indicate a degree of image sharpness in the image, each edge contributing a quantity to the focus signal, comprising:
 rejecting or deemphasizing, among a first plurality of edges whose first widths are same, any edge whose second width lies outside a first acceptance range, wherein each edge among said first plurality of edges has a gradient profile with a peak gradient, and said first width of said each edge is a width of said gradient profile at a first gradient level that is a first fraction of said peak gradient, wherein said second width of said any edge is another width of said gradient profile at a second gradient level that is a second fraction of said peak gradient, wherein said first and second fractions are common across said first plurality of edges.   
     
     
         15 . The method of  claim 14 , further comprising:
 rejecting or deemphasizing, among a second plurality of edges that all have a same third width across their respective gradient profiles at respective gradient levels that are at said first fraction of their respective peak gradients, any edge whose fourth width lies outside a second acceptance range,   wherein said third width and said second acceptance range are different from said first width and said first acceptance range respectively,   wherein said fourth width is a width across said gradient profile at a gradient level that is at said second fraction of said peak gradient of said any edge among said second plurality of edges.   
     
     
         16 . The method of  claim 14 , wherein said rejecting or deemphasizing is performed in an Edge Detection & Width Measurement Unit housed in a package together with a pixel array. 
     
     
         17 . The method of  claim 14 , wherein said rejecting or deemphasizing is performed in an Edge Detection & Width Measurement Unit housed in a package together with a processor that controls a focus position of a focus lens. 
     
     
         18 . The method of  claim 14 , wherein said rejecting or deemphasizing is performed in an Edge Detection & Width Measurement Unit housed in a package that does not contain a processor that controls storing of compressed images to a removable memory card. 
     
     
         19 . The method of  claim 14 , wherein said rejecting or deemphasizing involves executing computer-executable instructions. 
     
     
         20 . A non-transitory computer readable medium, comprising computer executable instructions and or parameter values for executing the method of  claim 14 . 
     
     
         21 . The method of  claim 14 , wherein said quantity is an edge width. 
     
     
         22 . The method of  claim 14 , wherein said quantity does not vary if all pixel values of the image that enter a computation of said quantity are scaled up by a common multiplier. 
     
     
         23 . The method of  claim 14 , wherein said quantity has a dimension based solely on a dimension of length. 
     
     
         24 . The method of  claim 15 , wherein all widths have been corrected for slants of the respective edges from their respective edge directions in which pixels of the respective edges are arrayed. 
     
     
         25 . The method of  claim 1 , wherein said quantity is an edge width. 
     
     
         26 . The method of  claim 1 , wherein said quantity does not vary if all pixel values of the image that enter a computation of said quantity are scaled up by a common multiplier. 
     
     
         27 . The method of  claim 1 , wherein said quantity has a dimension based solely on a dimension of length.