In the color desktop publishing system, in order to reproduce the image colors more realistically on the CRT, the color characteristics of the CRT need to be studied to achieve the conversion between the CRT color space and the standard color space XYZ. In general, when both conversions are implemented, it is assumed that:
1) Phosphorus powder has constancy, and the phosphorescence characteristics of phosphor powder do not change;
2) The space is uniform, and the color development characteristics of each location on the CRT are consistent;
3) Three electron guns are mutually independent and other conditions. However, these assumptions often do not hold, resulting in inaccurate conversion models. To solve this problem, many nonlinear transformation models are proposed. This paper analyzes and evaluates the transformation model.
One, Y correction method This method is divided into two steps, first realize the conversion between CRT phosphor brightness and electron gun voltage; and then realize the conversion between CRT phosphor brightness and XYZ tristimulus value.
1. The relationship between CRT phosphor brightness and electron gun voltage The relationship between CRT phosphor brightness and electron gun voltage is a nonlinear relationship. It is generally assumed to be an exponential relation. After analysis, it is found that when the input digital drive value DR=DG=DB=0 In theory, the tristimulus value measured on the above should be equal to zero, but the tristimulus value actually measured on a CRT is X=0.2327, Y=0.2201, and Z=0.2956. In order to eliminate this error, the tristimulus value measured at DR=DG=DB=0 is subtracted from the measured tristimulus value, and linear regression is then performed to improve the calculation accuracy.
Another method of solving is to use a look-up table method, that is, to establish a correspondence table of digital drive values ​​and luminance values ​​for each electron gun.
2. The conversion of the phosphor brightness values ​​LR, LG, LB and the tristimulus value XYZ.
Second, based on mathematical fitting method Linear transformation model The linear model is simple, only need to solve 12 coefficients, the calculation speed is faster, but the conversion accuracy is not high.
2. The polynomial model has a high conversion accuracy and is a more practical method. However, when there are color conversions with color gamut compression, the error is large.
3. The table lookup method stores the measured CRT RGB and CIE XYZ standard data in the form of a table. If RGB data is given during use, the corresponding XYZ data can be found from the table. Since R, G, B each component has 256 values, the size of the table after R, G, B combination is 256×256×256, and the table is very large. For this reason, the value of each component of R, G, B may be taken as a number of values, and the table will become smaller. For data that does not exist in the table, it can be obtained by interpolation.
Look-up table method is a very simple and effective method, but the establishment of the form requires testing a considerable amount of data, the workload is enormous, and the form needs to occupy a large amount of storage space.
First, the neural network method 1. Neural Network Method Introduction The neural network can complete the mapping of many nonlinear functions. It can be used to map the colors in one color space to another color space, such as RGB to XYZ, HVC to XYZ, and so on. In the neural network model, what is currently used is a multi-layer error back propagation network, ie, a BP network. It consists of an input layer, a hidden layer, and an output layer. In the conversion from the CR color space to the XYZ color space, the input layer consists of 3 nodes, which represent R, G, and B; the output layer also consists of 3 nodes, corresponding to X, Y, and Z; for the hidden layer, it needs to be actual Determine the number of hidden layers and the number of nodes.
Using neural network to realize CRT color space conversion is to learn the network according to the known input and output color data. Through the continuous learning of sample data, the weights of connecting each neuron are adjusted to make the converted color data closer to the real. The color data. For the BP network, when a network is given an input mode, it is transmitted from the input unit to the hidden layer unit, processed layer by layer by the hidden layer unit, and then sent to the output layer unit. After processing by the output layer unit, an output pattern is generated. This is a process of layer-by-layer status update called forward network. If there is an error between the output value and the expected value and does not meet the requirements, then it will be transferred to the error backward propagation process. The error value will be transmitted layer by layer along the connection path and the connection weights of the layers will be corrected.
2. Conversion Model This conversion model does not use a 3-input, 3-output network model. Therefore, the network training time can be greatly reduced, and the conversion accuracy can be improved.
1) Initialize the network learning parameters 2) Provide standard color test data, train the network until the color difference is satisfied br> 3) Forward propagation process: For a given color input data, calculate the network color data, And compare with the actual color data; if there is an error, execute 4; otherwise, return 2.
4) The error backward propagation process uses a neural network model, and the color conversion accuracy is high. However, it involves many problems, such as the determination of the number of hidden layers and the number of nodes, the setting of initial weights, and the selection of learning factors. These parameters must be determined through experiments.
Methods and Techniques for Injecting Ink into Integral Inkjet Heads Inkjet printers have gained popularity among home users for their inexpensive price and proximity to laser printers, but they have also been plagued by expensive supplies. At present, the manufacturers all integrate the ink jet printer's print head and ink. After the ink is used up, the print head can only be replaced together. Spending nearly 20 yuan, but can only print four or five hundred pages, which is a big burden for the average user, in fact, after the ink is used up, the life of the print head does not expire at all, can be fully refilled ink (inkjet printer dedicated Ink), continue to use, and print quality will not change, refilling the ink is only equivalent to a quarter of the cost of replacing the print head.
Home inkjet printers such as Canon, Epson and Hewlett-Packard can use a third-party manufacturer's fill ink, although the manufacturer stated in the instructions that the failure caused by the use of filled ink is not warranted, but in fact there is no risk If an inkjet head is used normally, there will be no problem with filling the ink three or four times.
When the ink has run out, carefully remove the printhead from the printer. Be careful not to touch the printhead's nozzle. Find a small hole on the top side of the print head, about 3 mm, with a cylindrical small plastic plug. Drill a small plastic plug with a small tap supplied with the fillable ink and pull the small plug aside. Do not morally, still have to plug in after pouring ink. Cut the long mouth of the SK-filled ink bottle into a small hole and insert it into the small head of the print head. Slowly inject about 20 ml of ink. Be careful not to spill it. After filling, plug the small plug and then reload the print head. The printer can be used normally. If you do not care about the ink overflow, you can eat it out of the stickers, under normal circumstances can be filled with ink 30,70,100 ml and so on, if you often print documents may wish to buy 100 millimeters long, relative to the family more.
There is also a problem to be noted that the ink jet head jams each time a new ink is inserted, regardless of how careful it is even if it does not touch the nozzle. At this time, do not use needles to tie the nozzle. This will completely damage the print head, because the nozzle is much smaller than we think, and it is very delicate. Since the ink jet printer is designed by using the principle that the ink is ejected from the small hole after being heated, when the nozzle is blocked, we can immerse the ink nozzle part in about 1 mm (a few seconds) in warm water. The heat will flush all blocked ink jets. After taking out the dining stickers, the water from the ink nozzles will be sucked from the side and the print heads will be reinstalled.
1) Phosphorus powder has constancy, and the phosphorescence characteristics of phosphor powder do not change;
2) The space is uniform, and the color development characteristics of each location on the CRT are consistent;
3) Three electron guns are mutually independent and other conditions. However, these assumptions often do not hold, resulting in inaccurate conversion models. To solve this problem, many nonlinear transformation models are proposed. This paper analyzes and evaluates the transformation model.
One, Y correction method This method is divided into two steps, first realize the conversion between CRT phosphor brightness and electron gun voltage; and then realize the conversion between CRT phosphor brightness and XYZ tristimulus value.
1. The relationship between CRT phosphor brightness and electron gun voltage The relationship between CRT phosphor brightness and electron gun voltage is a nonlinear relationship. It is generally assumed to be an exponential relation. After analysis, it is found that when the input digital drive value DR=DG=DB=0 In theory, the tristimulus value measured on the above should be equal to zero, but the tristimulus value actually measured on a CRT is X=0.2327, Y=0.2201, and Z=0.2956. In order to eliminate this error, the tristimulus value measured at DR=DG=DB=0 is subtracted from the measured tristimulus value, and linear regression is then performed to improve the calculation accuracy.
Another method of solving is to use a look-up table method, that is, to establish a correspondence table of digital drive values ​​and luminance values ​​for each electron gun.
2. The conversion of the phosphor brightness values ​​LR, LG, LB and the tristimulus value XYZ.
Second, based on mathematical fitting method Linear transformation model The linear model is simple, only need to solve 12 coefficients, the calculation speed is faster, but the conversion accuracy is not high.
2. The polynomial model has a high conversion accuracy and is a more practical method. However, when there are color conversions with color gamut compression, the error is large.
3. The table lookup method stores the measured CRT RGB and CIE XYZ standard data in the form of a table. If RGB data is given during use, the corresponding XYZ data can be found from the table. Since R, G, B each component has 256 values, the size of the table after R, G, B combination is 256×256×256, and the table is very large. For this reason, the value of each component of R, G, B may be taken as a number of values, and the table will become smaller. For data that does not exist in the table, it can be obtained by interpolation.
Look-up table method is a very simple and effective method, but the establishment of the form requires testing a considerable amount of data, the workload is enormous, and the form needs to occupy a large amount of storage space.
First, the neural network method 1. Neural Network Method Introduction The neural network can complete the mapping of many nonlinear functions. It can be used to map the colors in one color space to another color space, such as RGB to XYZ, HVC to XYZ, and so on. In the neural network model, what is currently used is a multi-layer error back propagation network, ie, a BP network. It consists of an input layer, a hidden layer, and an output layer. In the conversion from the CR color space to the XYZ color space, the input layer consists of 3 nodes, which represent R, G, and B; the output layer also consists of 3 nodes, corresponding to X, Y, and Z; for the hidden layer, it needs to be actual Determine the number of hidden layers and the number of nodes.
Using neural network to realize CRT color space conversion is to learn the network according to the known input and output color data. Through the continuous learning of sample data, the weights of connecting each neuron are adjusted to make the converted color data closer to the real. The color data. For the BP network, when a network is given an input mode, it is transmitted from the input unit to the hidden layer unit, processed layer by layer by the hidden layer unit, and then sent to the output layer unit. After processing by the output layer unit, an output pattern is generated. This is a process of layer-by-layer status update called forward network. If there is an error between the output value and the expected value and does not meet the requirements, then it will be transferred to the error backward propagation process. The error value will be transmitted layer by layer along the connection path and the connection weights of the layers will be corrected.
2. Conversion Model This conversion model does not use a 3-input, 3-output network model. Therefore, the network training time can be greatly reduced, and the conversion accuracy can be improved.
1) Initialize the network learning parameters 2) Provide standard color test data, train the network until the color difference is satisfied br> 3) Forward propagation process: For a given color input data, calculate the network color data, And compare with the actual color data; if there is an error, execute 4; otherwise, return 2.
4) The error backward propagation process uses a neural network model, and the color conversion accuracy is high. However, it involves many problems, such as the determination of the number of hidden layers and the number of nodes, the setting of initial weights, and the selection of learning factors. These parameters must be determined through experiments.
Methods and Techniques for Injecting Ink into Integral Inkjet Heads Inkjet printers have gained popularity among home users for their inexpensive price and proximity to laser printers, but they have also been plagued by expensive supplies. At present, the manufacturers all integrate the ink jet printer's print head and ink. After the ink is used up, the print head can only be replaced together. Spending nearly 20 yuan, but can only print four or five hundred pages, which is a big burden for the average user, in fact, after the ink is used up, the life of the print head does not expire at all, can be fully refilled ink (inkjet printer dedicated Ink), continue to use, and print quality will not change, refilling the ink is only equivalent to a quarter of the cost of replacing the print head.
Home inkjet printers such as Canon, Epson and Hewlett-Packard can use a third-party manufacturer's fill ink, although the manufacturer stated in the instructions that the failure caused by the use of filled ink is not warranted, but in fact there is no risk If an inkjet head is used normally, there will be no problem with filling the ink three or four times.
When the ink has run out, carefully remove the printhead from the printer. Be careful not to touch the printhead's nozzle. Find a small hole on the top side of the print head, about 3 mm, with a cylindrical small plastic plug. Drill a small plastic plug with a small tap supplied with the fillable ink and pull the small plug aside. Do not morally, still have to plug in after pouring ink. Cut the long mouth of the SK-filled ink bottle into a small hole and insert it into the small head of the print head. Slowly inject about 20 ml of ink. Be careful not to spill it. After filling, plug the small plug and then reload the print head. The printer can be used normally. If you do not care about the ink overflow, you can eat it out of the stickers, under normal circumstances can be filled with ink 30,70,100 ml and so on, if you often print documents may wish to buy 100 millimeters long, relative to the family more.
There is also a problem to be noted that the ink jet head jams each time a new ink is inserted, regardless of how careful it is even if it does not touch the nozzle. At this time, do not use needles to tie the nozzle. This will completely damage the print head, because the nozzle is much smaller than we think, and it is very delicate. Since the ink jet printer is designed by using the principle that the ink is ejected from the small hole after being heated, when the nozzle is blocked, we can immerse the ink nozzle part in about 1 mm (a few seconds) in warm water. The heat will flush all blocked ink jets. After taking out the dining stickers, the water from the ink nozzles will be sucked from the side and the print heads will be reinstalled.
Our factory stay on a place where is famous of knife manufacture .It is fabricated from high-quality stainless steel,single-piece, precision-stamped blade construction provides durability,professional, fine-edge bladeboasts precision cutting and is finely honed for long-lasting sharpness.We are good at Chef Knife,bread knife,carving knife,utility knife,paring knife,sharpener,kitchen scissor,steak knife.We can provide the Knife Set or single knife which up to you.
Household Knife,Dinner Knife,Household Floding Knife,Stainless Household Knife
YANG JIANG FEETON HOME PRODUCTS CO., LTD. , http://www.feetonbbq.com