1. A coefficient learning apparatus comprising:
student-image generation means for generating a student image, which has a quality degraded from the quality of a teacher image in accordance with a plurality of deterioration model equations each corresponding to one of a plurality of deterioration types, from said teacher image;
class classification means for sequentially setting each of pixels in said teacher image as a pixel of interest and generating a class for each pixel of interest from values of pixels in said student image located at positions corresponding to the position of each pixel of interest and peripheral positions surrounding said position of each pixel of interest;
weight computation means for computing a weight of each constraint condition equation of a plurality of constraint condition equations for at least some pixels in said teacher image based on the sum of feature quantities for said each class, said feature quantity representing a relation between said at least some pixels and adjacent pixels in said teacher image under each deterioration model equation used for finding the values of pixels in said student image from values of pixels located at positions corresponding to the position of each pixel of interest and peripheral positions surrounding said position of each pixel of interest; and
processing-coefficient generation means for generating a prediction coefficient to predict a value of each pixel of interest in said teacher image by carrying out a computation process applying to a plurality of pixels in the student image for each deterioration type and each class on the basis of a determinant including each deterioration model equation and each constraint condition equation.
2. The coefficient learning apparatus according to claim 1, wherein said weight computation means assigns a weight value of said constraint condition equation that is inversely proportional to a sum of feature quantities.
3. The coefficient learning apparatus according to claim 1, wherein said student-image generation means generates a student image by adding blurring appearances to said teacher image.
4. The coefficient learning apparatus according to claim 1, wherein said student-image generation means generates a student image by reducing said teacher image.
5. A coefficient learning method adopted by a coefficient learning apparatus, said coefficient learning method comprising:
generating a student image, which has a quality degraded from the quality of a teacher image in accordance with a plurality of deterioration model equations each corresponding to one of a plurality of deterioration types, from said teacher image;
sequentially setting each of pixels in said teacher image as a pixel of interest and generating a class for each pixel of interest from values of pixels in said student image located at positions corresponding to the position of each pixel of interest and peripheral positions surrounding said position of each pixel of interest;
computing a weight of each constraint condition equation of a plurality of constraint condition equations for at least some pixels in said teacher image based on the sum of feature quantities for said each class, said feature quantity representing a relation between said at least some pixels and adjacent pixels in said teacher image under each deterioration model equation used for finding the values of pixels in said student image from values of pixels located at positions corresponding to the position of each pixel of interest and peripheral positions surrounding said position of each pixel of interest; and
generating a prediction coefficient to predict the value of each pixel of interest in said teacher image by carrying out a computation process applying to a plurality of pixels in the student image for each deterioration type and each class on the basis of a determinant including each deterioration model equation and each constraint condition equation.
6. A non-transitory computer readable medium encoded with a coefficient learning program to be executed by a computer to carry out a coefficient learning process comprising:
generating a student image, which has a quality degraded from the quality of a teacher image in accordance with a plurality of deterioration model equations each corresponding to one of a plurality of deterioration types, from said teacher image;
sequentially setting each of pixels in said teacher image as a pixel of interest and generating a class for each pixel of interest from values of pixels in said student image located at positions corresponding to the position of each pixel of interest and peripheral positions surrounding said position of each pixel of interest;
computing a weight of each constraint condition equation for at least some pixels in said teacher image based on the sum of feature quantities for said each class, said feature quantity representing a relation between said at least some pixels and adjacent pixels in said teacher image under each deterioration model equation used for finding the values of pixels in said student image from values of pixels located at positions corresponding to the position of each pixel of interest and peripheral positions surrounding said position of each pixel of interest; and
generating a prediction coefficient to predict the value of each pixel of interest in said teacher image by carrying out a computation process applying to a plurality of pixels in the student image for said each deterioration type and each class on the basis of a determinant including said each deterioration model equation and each constraint condition equation.
7. A coefficient learning apparatus comprising:
a student-image generation section configured to generate a student image, which has a quality degraded from the quality of a teacher image in accordance with a plurality of deterioration model equations each corresponding to one of a plurality of deterioration types, from said teacher image;
a class classification section configured to sequentially set each of pixels in said teacher image as a pixel of interest and generate a class for each pixel of interest from values of pixels in said student image located at positions corresponding to the position of each pixel of interest and peripheral positions surrounding said position of each pixel of interest;
a weight computation section configured to compute a weight of each constraint condition equation of a plurality of constraint condition equations for at least some pixels in said teacher image based on the sum of feature quantities for said each class, said feature quantity representing a relation between said at least some pixels and adjacent pixels in said teacher image under each deterioration model equation used for finding the values of pixels in said student image from values of pixels located at positions corresponding to the position of each pixel of interest and peripheral positions surrounding said position of each pixel of interest; and
a processing-coefficient generation section configured to generate a prediction coefficient to predict the value of each pixel of interest in said teacher image by carrying out a computation process applying to a plurality of pixels in the student image for each deterioration type and each class on the basis of a determinant including each deterioration model equation and each constraint condition equation.
The claims below are in addition to those above.
All refrences to claim(s) which appear below refer to the numbering after this setence.
1. A process for the manufacture of a profiled wire of hydrogen-embrittlement-resistant, low-alloy carbon steel for flexible pipelines for the offshore oil and gas operations sector comprising:
providing a low-alloy carbon steel wire rod having a composition including, expressed in percentages by weight of the total mass:
75<C %<0.95;
30<Mn %<0.85;
Cr\u22660.4%;
V\u22660.16%; and
Si\u22661.40%,
the rest being iron and the inevitable impurities from smelting of the metal in the liquid state;
hot-rolling the wire rod in an austenitic region above 900\xb0 C.;
cooling the wire rod to ambient temperature;
subjecting the wire rod to isothermal quenching to obtain a homogeneous pearlitic microstructure;
subjecting the wire rod to an operation of cold mechanical transformation, carried out with a global work-hardening ratio of from approximately 50 to 80%, to give the wire rod a diameter of from approximately 5 to 30 mm;
subjecting the drawn wire to a short-duration recovery heat treatment carried out below an Ac 1 temperature of the steel.
2. The process for the manufacture of a profiled wire as recited in claim 1, wherein the short-duration recovery heat treatment is carried out at a temperature from 410 to 710\xb0 C. for a duration of one minute or less.
3. The process for the manufacture of a profiled wire as recited in claim 1, wherein the isothermal quenching is a patenting operation a molten lead bath.
4. The process for the manufacture of a profiled wire as recited in claim 3, wherein the patenting occurs at a constant temperature in a range from 520 to 600\xb0 C.
5. The process for the manufacture of a profiled wire as recited in claim 1, wherein the cold mechanical transformation includes drawing and cold rolling.
6. The process for the manufacture of a profiled wire as recited in claim 1, wherein the short-duration recovery heat treatment results in a mean tensile strength Rm of 1380 to 1920 MPa.
7. The process for the manufacture of a profiled wire as recited in claim 1, wherein the short-duration recovery heat treatment results in the profiled wire having a mean tensile strength Rm of 1380 to 1920 MPa.
8. The process for the manufacture of a profiled wire as recited in claim 1, wherein the short-duration recovery heat treatment results in the profiled wire having a mean yield strength Re of 1190 to 1730 MPa.
9. The process for the manufacture of a profiled wire as recited in claim 1, wherein the short-duration recovery heat treatment results in the profiled wire having a mean elongation at break from 9.6% to 12.0%.
10. The process for the manufacture of a profiled wire as recited in claim 1, wherein 1.4%\u2267Si\u22670.15%.
11. The process for the manufacture of a profiled wire as recited in claim 1, wherein the composition includes Al\u22660.06%.
12. The process for the manufacture of a profiled wire as recited in claim 1, wherein the composition includes Ni\u22660.1%.
13. The process for the manufacture of a profiled wire as recited in claim 1, wherein the composition includes Cu\u22660.1%.