TW476222B - Device and method for the simultaneous processing of global/local pixel image quality enhancement - Google Patents

Abstract

The present invention relates to a device and method for the simultaneous processing of global/local pixel image quality enhancement, especially relates to a device and method for the simultaneous processing of global/local pixel image brightness and contrast quality enhancement that alternatively applies a concept of global and local pixel image quality enhancement processing in an image. The main structure comprises a numerical conversion system to convert the intensity and brightness data of the three primaries red (R), green (G), and blue (B) of the input image, and convert the brightness value and the values on the log plane; an image processing system to simultaneously proceed the global digital processing and local pixel image quality enhancement processing for the value on the outputted log plane of the conversion system, and to send the processed value on the outputted log plane to the conversion system for reverse conversion before being outputted, whereby the quality of image directly inputted by sensor, whose details are lost due to focusing or exposure problem, can be improved. It is clearer in the brightness and contrast quality, and it is more fitted to the vision recognition of human brain.

Description

476222 _I ^. ： Ϋ_ V. Explanation of the invention (1) --------------- II "IIihi The present invention relates to a device capable of simultaneously processing global and local point image quality enhancement and its Method, especially using the concepts of global image processing and local point image quality enhancement processing at the same time, interactively applied to an image to achieve the purpose of improving the sharpness, brightness and contrast quality of image details. Due to the advancement of digital image processing technology and its wide application, many images from different sources can be digitized by computers to provide different applications. The most common method for processing images using a computer is to input the image recorded by the sensor into the computer through an input device. The input device converts the image into a digital format, that is, each pixel in the image is in three primary colors: red, green, and blue. The intensity value represents and is stored in the memory; the intensity value of each pixel's red, green, and blue represents its content in steps of 25.6, and provides various digital image processing purposes. However, the image data recorded by general sensors is far from our actual visual experience. The scene in front of the sensor contains light and dark parts. The sensor adjusts according to different exposures to capture different images. An image often has both underexposed shadows and overexposed. For different exposure adjustments, you can only adjust the exposure of the entire area of the frame to the average optimal value, but you must sacrifice the exposure quality of some parts of the frame. High technology has a huge negative impact. Furthermore, the sensor itself has a hardware design problem. Different hardware designs will have different sensitivities in the spectral bands. Technically, the sensor

Page 5 476222 V. Description of the invention (2) The detector can only be based on the sensitivity of the user. In addition, in the process of digital limit and quantization limit processing, the value range and, in some cases, the brightness is very different from that of the original brain vision image processing, which is called dynamic compression. The adjustment of the focal length of the sensing sensor, focusing only, caused some parts of the sensor to be captured, but because of the depth of field, the distortion of the image affected the human vision. The focus is mainly on the three-dimensional search of the local lens. The quality of the light image with the perforated hole can meet the vision. Therefore, when compared to the three balls, after such a visual mode color or brightness change, the overall clear details and connotations, especially the differences are more obvious. Press, Learn Digital Imaging technology does not have a set of "correction at the same time": "I ..-^ t ^ er requirements design, the recording unit with the best sensor on a certain band in addition to the range of the system, each The limitation of digital analog quantization of an analog signal will cause a loss of signal accuracy, which will cause a significant difference in the recorded digital. This problem is compressed in the digital shadow value range (Dyn amic Range

The device is facing a three-dimensional world. On the image that can be aligned for a certain theme in the daytime, some parts are clear and problematic, and the blurry noise caused by the inaccuracy of focus is even worse. As obvious. It is a subject-based search method, which adjusts the amount of eyeballs relative to the signal of the eyeballs reflected in each domain, so that each point in space takes the best value in experience. Each point of the signal in the degree space enters the eye type from the image. The image reconstructed in the brain, whether it is a face or a face, has more parts than the image recorded from the sensor in the shadow and strong light. The aforementioned sensor quality issues address exposure, focus, and dynamic range compression issues.

Page 6 476222

The beginning of the digital image is three primes. The global correction in this pixel can only be very long because the amount of the image has reached the entire image. Because of the amount of the image, the original (B) line is considered to be the same as that of the entire image. Simultaneously with the moving image leveling, the whole set of (R) is recorded at the same time as the adjustment of the whole world, and the image of the original setting is repaired and then the element is processed. There are fewer problems. Some methods are still impossible, and). Blue time π digital video image management system, color red (R correction is the first green pigment into the positive, the number of parts to improve the integration of the image and the loss of the overall special image, the intensity value of the system does not meet the effect The processed material is transferred to the place where the image is recorded), green (the image detail of the pixel in the pixel's global trimming image, and the original image brightness is based on the brightness of the original image. When the pixel is processed in detail, it is based on the principle of recovery. The system is G), the blue-red pigment is positive, and the exposure in this last digit is brighter than the original segment; A then produces the quality of the flat segment. The three primary colors are red. The intensity of the first digital shadow in the unit recorded by the detector is sequentially processed. Correction: The blue color processing system in the element is uniform in the range of the state, and the correction effect of the correction is reduced. The green (G type) costs the latest digital image processing method, which is processed in the image processing system. The method of the intensity values of the three primary colors of pixels (R), green (G), and blue (B) is changed from global correction to local point correction. In each primary color map, global correction is performed with each region as a unit. That is, in each area = no focus is determined, and the relative relative processing is performed in accordance with the focus. The smaller the area is, the higher the sharpness of the contrast of the image. However, the image processed by this method has the three primary colors of pixels processed separately. The color vector that is recombined will produce a deviation that can be clearly recognized by the naked eye in the color coordinate space, and loses the original 476222 90.! 2 · 1 8 labor: a.:, .1. ': 卜',.} 丨： V. Explanation of the invention (4) "---------------------" There is a color in visual experience, so it is necessary to introduce another adjustment value for color restoration adjustment. Can be as close as possible to the color category in the visual experience. At the same time When performing local point correction, it also adopts the method of correcting and combining on each primary color map separately, which also takes an extremely long calculation time to complete the processing once; and in this way, only a logarithmic brightness conversion is used Mode, trying to solve the overall adjustment of exposure brightness and dynamic range compression, it is obvious that for some overexposed image parts, such processing cannot achieve the best visual adjustment. To this end, the present invention develops a set Algorithms, simulating the rules of visual experience, re-correcting the image signals recorded by the sensor, and simultaneously solving the problems of exposure, focus, and dynamic range compression, in order to pursue the highest realm that conforms to the actual human visual experience. Therefore, how to establish an effective image enhancement processing device and method, not only will not cause the deviation of the color vector of the three primary colors of pixels in the color coordinate space during the image enhancement processing, but the processing speed will be faster, the image will be clearer, and the dynamics will be considered. The best adjustment of the range compression, exposure and focus is really the user ’s longing hope and the difficulties that the inventor wants to solve. The inventor has developed a method that can simultaneously process the global and local points. The image augmentation device and its algorithm are improved for the lack of the above image augmentation processing methods, while solving the problems of exposure, focus, and dynamic range compression to achieve the purpose of making digital image quality better. That is, the main object of the present invention is to provide a device and method capable of simultaneously processing global and local point image quality enhancement, so as to increase the detail of the image in dark and bright places, thereby enhancing the real image. Quality.

Page 8

V. Description of the invention (5) Qiu Mingming's secondary project # 'is to provide-a device that can increase and display the image of the point and point of view, and to simultaneously process the global and "exposure status and overall dynamics: an image The content of the site adjustment is built to improve the image recording. Another object of the present invention is to provide a device and method that can increase and display the image, from $ to [Day]: J: The domain and the degree of capture can lead to significant savings. Signal processing time. "Effective U trial? The members can better understand the purpose, features and functions of the present invention, and explain the present invention in detail through the following examples and illustrations. The explanation is as follows: Please refer to FIG. 1 for the first month. Structure block diagram of the preferred embodiment 丨 As shown in the figure, the present invention is a device and method capable of simultaneously processing global and local point image augmentation and display. Its main structure includes a numerical conversion system 1 2, The numerical conversion system 12 converts the three primary color red, green, and blue intensity values of the digital image pixels of the input original image i 〇 into luminance values represented by brightness (Illuminati) and color coordinates, and The brightness value is converted to a logarithmic plane value; conversely, it can also be

A logarithmic plane value after global correction and local point correction is converted into a luminance value represented by luminance and color coordinates, and then the luminance value is converted into the intensity values of the three primary colors of the digital image pixels. An image processing system 14 includes a global correction unit 14 0 and a local point correction unit 14 2. The global correction unit 1 4 0 performs the calculation from the logarithmic plane of the numerical conversion system 12 Global correction, making the overall brightness of the image look more average than the original image, and reducing some lost images due to quantization

Page 9 5. Description of the invention Like details. Change system 1 The higher the logarithmic level is, the higher the local point is, the higher it becomes; 2 is for the subsequent transfer to the data processing and the shadows please refer to the picture; the intensity value color coordinate table in the figure is converted to the opposite number 2 3, the system Compared with the original shadow, another value is a local point and surface. It is divided into several areas. In each area, the positive value of the regional division and the numerical value of the local correction point singular plane are divided into several areas. Zone processing, so after the area correction unit 1 4 ', the conversion system 12 will be repaired. The strong element of the three primary colors 1 4 2 will be taken from this value for local point correction, which is also the area, with the unit of the area as a unit. Set a focal point in the domain. With the finer of the focal point, the image contrast sharpness 0 and the local point correction unit 1 ^ the numerical synthesis of the positive logarithmic plane 'in the numerical conversion system 12 within the degree value, so you can To get a better picture, first enter the number of 20 points on the value of 20 degrees, and reduce the point of the correction part to make it finer with each focus. The correction system

(6) The opposite number of Board 2 is corrected. Relative to the global step, the value is converted into an image like the one shown in Figure 2, as shown in Figure 2. As shown in the figure, the plane value of the number is divided into a focal image to average the number of the present invention; after a 'such as the step value, such as processing ， Remove some parts in the whole area. Correct. The area is points. The image pair is synchronized. A better image is color. Step 2 1 Step 2 is full I ”so that the amount is as the steps are about to come—the unit and the focus ratio are clear. In the embodiment, the original three conversions are performed; then 2; the field correction image is normalized and streamed 2, and the logarithm is performed at the full point, and then the image is red, so as not to cross the part in a pair. , The volume loss of the film is the same as the field correction part of the plane; the two repair process of the green, blueness and color number conversion logarithmically as steps look like details logarithmic value area 'First after the global correction process Of _

476222

V. Description of the invention (7) Numerical synthesis, such as the value of the color coordinate table and the intensity value of the color, such as 2 8; so of course, for the accuracy loss of the brightness numerical image, please refer to the value on the positive processing flow. Pass 2 4 2 and low-pass to the three sample values of low-level signals and input 0 and 2 4 4 0 If the abnormality is generally positive, then W-weighted 2 4 4 4 is used to adjust the internal scale 2 4 number), so that step 2 5 is completed, and then through the logarithmic clock 仏 七 gt ▲ ▲ change the number of mattresses to the brightness, as in step 2 6; then the brightness is the digital image pixel three primary colors red, green And blue V = 7; finally, it is output through the output image, and the whole image processing operation flow is completed in one step. In the above step 22, it is also possible to directly divide into two parts without performing the logarithmic conversion, but processing the digits on the logarithmic plane has less loss than processing directly with the brightness value. Figure 3 is a block diagram of local point modification in a preferred embodiment of the present invention; as shown in the figure, the low-pass filtering (I) 2 4 0, low-pass filtering) input to the logarithmic plane is first filtered (E) 2 4 4 'that is, by inputting a set of signals', it is mainly to take out the low-order signals that are not needed for the quality enhancement display from the values on the logarithmic plane, and subtract the values from the logarithmic plane. 2400, 242 ^, so that the low-order signal can be filtered out, and then the results are filtered 2402, 2422, and 2442. The main image data range is used as the basis. Except for the abnormal part, the weight is 2404, and the weight is 2424 and wg. 'In order to highlight the characteristics of a part, 2 4 6 will be synthesized through the weighted highlights, and the synthesized value is finally _τ · 8 (where 7 祎 is a local point correction scale. The value formed is a local point. Bit correction value.> It is worth mentioning that the main spirit of the present invention (image enhancement processing system

476222 90.12. 18 years :, 5. Description of the invention (8) system) can be expressed by the following equation η

I rc ΊϊΤΓΤυ W: ·)) LPF ^ I) where I is the brightness of the input and I 'is the brightness of the output; k is a constant; r is a parameter for global correction, and r is a scale parameter for local point correction. It can be adjusted automatically or according to the user's needs; w is a weighting parameter, which is 1 in total; LPF is a low-pass filter. Therefore, in the present invention, the image is mainly processed by local point correction and global correction. Before processing the image, the intensity values of the three primary colors of the digital image pixels are converted into brightness values expressed by brightness and color coordinates, and then the brightness values are converted. Logarithmic conversion, local point correction of the values on the logarithmic plane after the conversion, instead of individual point correction of the intensity values of the three primary colors, and global correction at the same time, which not only saves a lot of time, but also eliminates the need for It is an invention that is superior to the conventional technique to perform other correction steps. To sum up, the invention can not only shorten the working time, but also make the processed image clearer and better in terms of brightness and contrast, and it will not be plagued by color deviation. It is novel and progressive. And those who can be used in the industry should comply with the patent application requirements of China's patent law. No doubt, the invention patent application was submitted in accordance with the law. However, the above is only a preferred embodiment of the present invention, and is not intended to limit the scope of implementation of the present invention. For example, all methods, structures, features, and spirits described in the scope of patent application for the present invention are equally changed and Embellish

Page 12 80. 12, ί 8. 476222 V. Description of the invention (9)

All should be included in the scope of patent application of the present invention. 0 Brief description of drawing number • 1 0 Original image 1 2 Numerical conversion system 1 4 Image processing system 1 4 0 Global correction mouth early 1 4 2 Local point correction unit 1 6 Processed image 2 0 Into image 2 1 Color conversion 2 2 Logarithmic conversion 2 3 Global correction 2 4 Local point correction 2 4 0 Low-pass filtering (I) 2 4 0 0 Subtraction 2 4 0 2 Abnormal filtering 2 4 0 4 W 1 Weighted 2 4 2 Low-pass filtering (Π) 2 4 2 0 Subtraction 2 4 2 2 Abnormal filtering 2 4 2 4 W 2 Weighted 2 4 4 Low-pass chirp (Π) 2 4 4 0 Subtraction 2 4 4 2 Anomaly filtering, 2 4 4 4 W 3 Weighted 2 4 6 Internal synthesis 2 4 8 Scale adjustment 2 5 Out of the image Page 13 476222 π "

Page 14

Claims (1)

III 6. Scope of patent application 2 4 5 6 1 · A type that can simultaneously process all the essential structures includes at least one numerical conversion system, the conversion between special values, an image processing system, and the point correction belongs to the following: The unit then performs the digital correction action and converts the intensity value of the primary color at the point position of the patent application range to the conversion system. The value of the dress. For example, the image at the point of the patent application is displayed on the number plane ^. For example, the application of the first point image display area and local point image augmentation display device in the scope of the patent application can be used for: • and the system will be converted to 1 when the global value conversion of the global line is included. 2 laboratories, including 1 laboratories on the plane, in which the intensity values of the three primary colors of pixels, a global correction unit and a local ^ * \ correction unit are converted from this value to a corrective action, and the local point repair system is introduced The numerical values are synthesized after the local points are positive and returned to the above-mentioned, which can simultaneously process the global and local values. The numerical conversion system is a three-dimensional pixel value. It can be described that it can process both global and local. The numerical conversion system is a numerical value that can be turned on again. The special value that can process both the global and the partial is the one described in the item 1 which can process the global and the partial at the same time, where the special value that is passed in is a pair of 1 that can handle the entire field simultaneously And local, where the global amendment is borrowed from a standard reference Page 15 476222, Proclaimed Strange Shoulder L Spear ij Series Color Circumference 1'—1 — ——— β—] The test table specifies the r parameter to perform this global correction. 7 · As described in item 6 of the scope of the patent application, the global and local point image display device can be processed simultaneously, where the τ parameter is one of automatic setting and manual setting. 8 · A method that can simultaneously process global and local point image enhancement. The main steps include at least: a. Converting the image to be processed to the intensity value of the three primary colors of the pixel 9 b. The intensity value of the three primary colors of the pixel Convert it into a brightness value; c. Simultaneously perform global correction and local point display correction on the brightness value d · convert the result value in step c into the intensity value of the three primary colors of the pixel; and e. Send the intensity value of the three primary colors of the pixel To the display. 9 · As described in item 8 of the scope of the patent application, the global and local point image enhancement method can be processed at the same time. In step c, the global correction is performed in accordance with the parameters specified in the standard reference table for the global correction. I 0 · As described in item 9 of the scope of the patent application, the global and local point image display method can be processed simultaneously, where the r parameter is one of automatic setting and manual setting. II · As described in item 8 of the scope of the patent application, it is possible to simultaneously process the global and local point image enhancement methods. In step c, the steps of the local point enhancement display correction are: c 1. Make the brightness value low Pass filtering Page 16 Miscellaneous 476222 VI. Patent application scope one ... one c 2. Subtract the brightness value from the result value in step c 1 to obtain a subtraction value; c 3. Filter the abnormal part of the subtraction value , And then multiply one weighting parameter Wi, c 4. Add all the values obtained in step c 3; c 5. Adjust the values obtained in step c 4 by an r scale to obtain a numerical value of local point increase and correction. ; And c 6. Add the result value in step c 5 and the result of the global correction 〇 1 2 · As shown in item 11 of the scope of patent application, the global and local point images can be processed simultaneously , Where the low-pass filtering system in step c 1 can be divided into 1, 2 · · · η total N groups. 1 3 · The method for simultaneously processing global and local point image enhancement as described in item 11 of the scope of patent application, wherein the weighting parameter W in step c 3 is from Wp W2 · · · Ν group. 14 · The method for processing global and local point image enhancement simultaneously as described in item 13 of the scope of patent application, wherein the weighting parameter W sums up to 1. 15 · The global and local point image enhancement method as described in item 11 of the scope of patent application, wherein in step c5, the r-son is a scale adjustment parameter and can be selected from a standard reference table. 16 · As described in item 15 of the scope of patent application, the global and local point image display method can be processed at the same time, where the 7 parameters are optional one of automatic setting and manual setting. 1 7 · Can handle global and local parts simultaneously as described in item 8 of the scope of patent application Page 17 476222 90 ″ 2 · 18 Amendment VI. Method of patent application TJfjytj point image display method, where step C can also be replaced by the following steps: c 1. Convert the brightness value to a value on the logarithmic plane; c 2 . Performing global correction and local point-increment correction on the values on the logarithmic plane at the same time; and c3. Converting the result value in step d to a luminance value. Page 18