JP2007158388A5

JP2007158388A5 -

Info

Publication number: JP2007158388A5
Application number: JP2005346416A
Authority: JP
Filing date: 2005-11-30
Publication date: 2009-01-22
Anticipated expiration: 2025-11-30

Claims

The R, G, B values digitized by the sensor are converted into V values representing brightness, C _RG values representing the color difference between R and G values, and C _YB values representing the color difference between Y and B values. As for the type, the correspondence between the input value and the output value uses a function that draws an inverted S-shaped curve to correct the brightness of dark and bright areas, and corrects the image. The amount of correction is determined from the overall domain of the image and the local domain of the image, and in the C _RG and C _YB values for color contrast, the correspondence of the output value to the input value draws an S-shaped curve By using each of these functions, the contrast between the R value and G value, the contrast between the Y value and B value is increased, and the correction for increasing the visibility of the color is performed. determining a correction amount from the local variable area, V values, C _RG value, corrects the C _YB values independently Image quality improving method characterized by integrating normalized pixel value last.

(a) an input mechanism for inputting image information; (b) a V input mechanism for inputting a V value representing luminance from a value obtained by the input image; and (c) a value obtained by the input image. RG input mechanism that inputs C _RG value, which is the color difference between R value and G value, and (d) YB input that inputs C _YB value, which is the color difference between Y value and B value, from the value obtained from the input image A mechanism, and (e) a V-neighbor image extraction mechanism that receives an image from the V-input mechanism and extracts a pixel value of a neighboring V including the pixel for each pixel, and (f) receives an image from the RG input mechanism, An RG neighborhood image extraction mechanism that extracts pixel values of neighboring C _RG including the pixel for each pixel, and (g) a neighboring _CYB pixel including the pixel for each pixel that receives an image from the YB input mechanism. A YB neighborhood image extraction mechanism for extracting values; and (h) receiving a neighborhood image from the V neighborhood image extraction mechanism and measuring a local region of V near the pixel. And V neighboring image pixel value variance range analysis mechanism, (i) receiving said neighboring images from RG neighboring images extracted mechanism, C _RG RG neighboring image pixel value variance range analysis to measure the local domain of the value of the pixel neighborhood mechanism and, (j) the YB receive neighboring images from the vicinity extraction mechanism, and YB neighboring image pixel value variance range analysis mechanism for measuring the local domain of C _YB value of the pixel neighborhood, (k) the V input mechanism Based on the result of the V neighborhood image pixel value domain analysis mechanism, the correction amount of the pixel value at the center of the V neighborhood is determined by the lower limit value of the image, the standard deviation value, the average value of the entire image, and the average value of the neighborhood of the image V _on-center image pixel value variable determination mechanism, (l) based on the results of the V input mechanism and the V neighborhood image pixel value domain analysis mechanism, the upper limit value of the image, the standard deviation value, the average value of the entire image V _Off-center image image that determines the correction amount of the pixel value at the center of the V neighborhood by the average value of the neighborhood of the image A pixel value variable determining mechanism, (m) the V _On-center by the variable obtained by the image pixel value variable determining mechanism and V _Off-center image pixel value variable determining mechanism, said V _On-center image pixel value and V A V pixel value correction mechanism that determines the pixel value of the V-neighbor center pixel with a function that draws an inverse S-shaped curve, as obtained by correcting the _Off-center image pixel value and obtaining the logarithm of these ratios, (n) V pixel value normalization mechanism that normalizes the pixel value obtained by the V pixel value correction mechanism, and (o) based on the results of the RG input mechanism and the RG neighborhood image pixel value domain analysis mechanism, An RG pixel value correction mechanism that determines the overall correction amount with a curve that draws an S-shape such as a sigmoid function, and (p) RG pixel value normalization that normalizes pixel values obtained by the RG pixel value correction mechanism And (q) a correction variable R _{On + Off} of R determined by a variable obtained from the RG neighborhood image extraction mechanism, and obtained by the RG normalization mechanism R neighborhood image pixel value adjustment mechanism that finely adjusts the value, and (r) G correction variable G _{On + Off} based on the G neighborhood center pixel value and neighborhood R average value obtained from the RG neighborhood image extraction mechanism Determining the G value obtained by the RG pixel value normalization mechanism G fine adjustment image pixel value adjustment mechanism, (s) R obtained by the R pixel value adjustment mechanism and the G pixel value adjustment mechanism A C _rg conversion mechanism that converts the pixel values of G and G into color differences between the R and G values, and (t) a sigmoid function based on the results of the YB input mechanism and the YB neighborhood image pixel value domain analysis mechanism. A YB pixel value correction mechanism that determines an overall correction amount with an S-shaped curve, and (u) a YB pixel value normalization mechanism that normalizes pixel values obtained by the YB pixel value correction mechanism, v) Based on the Y average between the locals including the pixel value of the Y neighborhood center obtained from the YB neighborhood image extraction mechanism and the B average between the locals including the pixel value of the B neighborhood center, the B value Determining a positive variable amount B _On, and B neighboring elementary image pixel value adjusting mechanism for adjusting the B value obtained from the YB pixel value normalizer, (u) Y vicinity obtained from the YB neighboring images extracted mechanism Based on the Y average between the locals including the central pixel value and the B average between the locals including the central pixel value in the vicinity of B, the correction variable amount Y _Off of the Y value is determined and obtained from the YB pixel value normalization mechanism. Y neighboring elementary image pixel value adjusting mechanism for finely adjusting the Y value, and (w) From the pixel values of Y and B values obtained by the B neighboring image elementary value determining mechanism and the Y neighboring image pixel value determining mechanism, and C _yb conversion mechanism for converting the color difference of the Y and B values, (x) the pixel value of C _rg obtained by the pixel value and C _rg conversion mechanism of V obtained in V pixel value normalizer and C _yb conversion The rgb conversion mechanism that integrates the _Cyb pixel values obtained by the mechanism and obtains a corrected rgb value; (y) Receives the pixel values of each pixel from the previous rgb conversion mechanism, composes and outputs the entire image Out An image quality improving apparatus comprising a force mechanism.

(a) an input mechanism for inputting image information; (b) a V input mechanism for inputting a V value representing luminance from a value obtained by the input image; and (c) a value obtained by the input image. RG input mechanism that inputs C _RG value, which is the color difference between R value and G value, and (d) YB input that inputs C _YB value, which is the color difference between Y value and B value, from the value obtained from the input image A mechanism, and (e) a V-neighbor image extraction mechanism that receives an image from the V-input mechanism and extracts a pixel value of a neighboring V including the pixel for each pixel, and (f) receives an image from the RG input mechanism, An RG neighborhood image extraction mechanism that extracts pixel values of neighboring C _RG including the pixel for each pixel, and (g) a neighboring _CYB pixel including the pixel for each pixel that receives an image from the YB input mechanism. A YB neighborhood image extraction mechanism for extracting values; and (h) receiving a neighborhood image from the V neighborhood image extraction mechanism and measuring a local region of V near the pixel. And V neighboring image pixel value variance range analysis mechanism, (i) receiving said neighboring images from RG neighboring images extracted mechanism, C _RG RG neighboring image pixel value variance range analysis to measure the local domain of the value of the pixel neighborhood mechanism and, (j) the YB receive neighboring images from the vicinity extraction mechanism, and YB neighboring image pixel value variance range analysis mechanism for measuring the local domain of C _YB value of the pixel neighborhood, (k) the V input mechanism Based on the result of the V neighborhood image pixel value domain analysis mechanism, the correction amount of the pixel value at the center of the V neighborhood is determined by the lower limit value of the image, the standard deviation value, the average value of the entire image, and the average value of the neighborhood of the image V _on-center image pixel value variable determination mechanism, (l) based on the results of the V input mechanism and the V neighborhood image pixel value domain analysis mechanism, the upper limit value of the image, the standard deviation value, the average value of the entire image V _Off-center image image that determines the correction amount of the pixel value at the center of the V neighborhood by the average value of the neighborhood of the image A pixel value variable determining mechanism, (m) the V _On-center by the variable obtained by the image pixel value variable determining mechanism and V _Off-center image pixel value variable determining mechanism, said V _On-center image pixel value and V A V pixel value correction mechanism that determines the pixel value of the V-neighbor center pixel with a function that draws an inverse S-shaped curve, as obtained by correcting the _Off-center image pixel value and obtaining the logarithm of these ratios, (n) V pixel value normalization mechanism that normalizes the pixel value obtained by the V pixel value correction mechanism, and (o) based on the results of the RG input mechanism and the RG neighborhood image pixel value domain analysis mechanism, An RG pixel value correction mechanism that determines an overall correction amount with a curve that draws an S-shape such as a sigmoid function, and (p) an RG pixel value normalization that normalizes pixel values obtained by the RG pixel value correction mechanism And (q) a correction variable R _{On + Off} of R determined by a variable obtained from the RG neighborhood image extraction mechanism, and obtained by the RG normalization mechanism R neighborhood image pixel value adjustment mechanism that finely adjusts the value, and (r) G correction variable G _{On + Off} based on the G neighborhood center pixel value and neighborhood R average value obtained from the RG neighborhood image extraction mechanism Determining the G value obtained by the RG pixel value normalization mechanism G fine adjustment image pixel value adjustment mechanism, (s) R obtained by the R pixel value adjustment mechanism and the G pixel value adjustment mechanism A C _rg conversion mechanism that converts the pixel values of G and G into color differences between the R and G values, and (t) based on the results of the YB input mechanism and the YB neighborhood image pixel value domain analysis mechanism, such as a sigmoid function A YB pixel value correction mechanism that determines an overall correction amount with an S-shaped curve, and (u) a YB pixel value normalization mechanism that normalizes pixel values obtained by the YB pixel value correction mechanism, v) Based on the Y average between the locals including the pixel value of the Y neighborhood center and the B average between the locals including the pixel value of the B neighborhood center obtained from the YB neighborhood image extraction mechanism, Determining a positive variable amount B _On, and B neighboring elementary image pixel value adjusting mechanism for adjusting the B value obtained from the YB pixel value normalizer, (u) Y vicinity obtained from the YB neighboring images extracted mechanism Based on the Y average between the locals including the central pixel value and the B average between the locals including the central pixel value in the vicinity of B, the correction variable amount Y _Off of the Y value is determined and obtained from the YB pixel value normalization mechanism. Y neighboring elementary image pixel value adjusting mechanism for finely adjusting the Y value, and (w) From the pixel values of Y and B values obtained by the B neighboring image elementary value determining mechanism and the Y neighboring image pixel value determining mechanism, and C _yb conversion mechanism for converting the color difference of the Y and B values, (x) the pixel value of C _rg obtained by the pixel value and C _rg conversion mechanism of V obtained in V pixel value normalizer and C _yb conversion The rgb conversion mechanism that integrates the _Cyb pixel values obtained by the mechanism and obtains a corrected rgb value; (y) Receives the pixel values of each pixel from the previous rgb conversion mechanism, composes and outputs the entire image Out Software characterized by having a force mechanism.