System for natural language assessment of relative color quality
A natural language and color technology, applied in the field of image quality analysis, which can solve the problems of interpretation experts, time-consuming, and difficult output of MICAM.
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
example 1
[0062] Angle: = 13
[0063] Quantity: = Red.12
[0064] Reference value: = red
[0065] Relative magnitude:= magnitude / reference magnitude
[0066] Relative magnitude = 0.12.
[0067] For example, if it is determined that the hue difference angle is 13 (which would fall within "red" n from Table 1), then the opposite hue color falls within "Green Blue" from Table 2. Additionally a magnitude of .12 was determined as shown in the above formula. Diagram this process as Figure 4 440 in.
[0068] The normalized magnitude of the color change is then passed from the magnitude determiner 120 to the natural language selector 130 to select the appropriate quantitative superlative word or phrase from the natural language table 131 according to a set of category quantization thresholds for hue and opposite hue. Picture this as Figure 4 Process 450.
[0069] An example natural language table 131 for a particular color shift threshold is illustrated in Table 3.
[0070] Table 3 (...
example 2
[0091] If the following values are input to the saturation measurement facility 114:
[0092] meanAbsTest_a:=.1 meanAbsRef_a:=.11
[0093] meanAbsTest_b:=.08 meanAbsRef_b:=.1
[0094] Then diffOfMeanAbs_a:=meanAbsTest_a - meanAbsRef_a = -0.01
[0095] diffOfMeanAbs_b:=meanAbsTest_b - meanAbsRef_b = -0.02
[0096] First as described as Figure 5 As in process 510 of , the distance between the test and reference image(s) is determined as given above. Quantization bins for the corresponding distances are then determined in process 520, which means determining which index value corresponds to each diffOfMeanAbs_a and diffOfMeanAbs_b. If two distances are quantized to the same bin (same threshold level of Table 3), then image 3 The natural language selector 130 can be as in Figure 5 The processes shown in 530 and 540 generate the following text for describing certain conditions:
[0097] All in all, the video colors appear to be {superlative 1} {more saturated / less s...
example 3
[0101] If the following values are input to the saturation measurement facility 114:
[0102] meanAbsTest_a:=.3 meanAbsRef_a:=.11
[0103] meanAbsTest_b:=.08 meanAbsRef_b:=.1
[0104] Then diffOfMeanAbs_a:=meanAbsTest_a - meanAbsRef_a = 0.19
[0105] diffOfMeanAbs_b:=meanAbsTest_b - meanAbsRef_b = -0.02
[0106] Here, unlike Example 2, when the quantization bins for "a" and "b" are different according to Table 3, image 3 The Natural Language Selector 130 can generate the following text:
[0107] In summary, the video has {superlative 1} {more saturated / less saturated} red and / or green and {superlative 2} {more saturated / less saturated} blue and / or yellow.
[0108] This becomes, using the data from Example 3:
[0109] "Overall, the video has noticeably more saturated reds and / or greens and slightly less saturated blues and / or yellows."
[0110] Overall change in color variety / variation
[0111] Color Diversity / Change Measurement Facility 116 ( figure 2 ) Deter...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 