TWI407383B

TWI407383B - Image quality enhancement method

Info

Publication number: TWI407383B
Application number: TW97134464A
Authority: TW
Original assignee: Univ Nat United
Priority date: 2008-09-08
Filing date: 2008-09-08
Publication date: 2013-09-01
Also published as: TW201011692A

Abstract

This invention provides an image quality enhancement method for using to enhance the quality of grayscale and color images. The quality enhancement method for grayscale image is to acquire in advance the original grayscale image of the image of which the quality is to be enhanced, and then to sample the pixels of the original grayscale image with predetermined sampling rate so as to create a sampled image, which is then treated with equalization process to create a sampled and equalized image, and after that, a look-up table is created according to the sampled and equalized image; finally the original grayscale image is transferred to a new grayscale image with enhanced quality through the look-up table. The quality enhancement method for color image is to acquire in advance the original color image of the image of which the quality is to be enhanced, the color space of original color image is converted to a color space containing brightness Y, and then to sample the brightness Y of the original color image to create a sampled brightness with predetermined sampling rate, which is then treated with equalization process to create a sampled and equalized brightness. After that, a look-up table is created according to the sampled and equalized brightness, and the brightness Y of original color image is transferred to a new brightness of color image with enhanced quality through the look-up table. Finally, the color space of enhanced color image with new brightness is converted to original color space.

Description

影像品質強化之方法Image quality enhancement method

本發明係與影像品質改善之技術領域有關，更詳而言之是指一種可提昇運算速度並縮小硬體代價之影像品質強化之方法者。The present invention relates to the technical field of image quality improvement, and more specifically to a method for enhancing the image quality and improving the image quality of the hardware.

按，有時影像會因對比(contrast)不佳之因素看起來模糊不清，甚至造成識別上的困難，對於監看環境以維護公共安全之監視系統而言，顯然會影響監視效果之確實與否。為提高影像之對比程度，目前有不少方法可資運用，其中，直方圖等化法(histogram equalization)是簡單而有效之技術，其原理是藉由重新分配影像中的灰階分佈，使其成為均勻分佈來強化明暗對比，而得到較佳的顯示狀態，如中華民國發明第92116702號「以區域方式偵測的對比強化(CONTRAST ENHANCEMENT)方法」、第92116703號「使用滑動視窗方式的對比強化(CONTRAST ENHANCEMENT)方法」、第93107846號「圖像對比調整方法」、第93113506號「利用小規模資料矯正不均勻影像外觀之技術」、第94127524號「影像調整方法與裝置」及第94132726號「影像邊緣強化裝置及其方法」等專利案所示。Press, sometimes the image may be blurred due to poor contrast, and even cause difficulty in identification. For monitoring systems that monitor the environment to maintain public safety, it will obviously affect the accuracy of the monitoring effect. . In order to improve the contrast of images, there are many methods available. Among them, histogram equalization is a simple and effective technique, which is based on redistributing the gray scale distribution in the image. It becomes uniform distribution to enhance the contrast between light and dark, and obtains a better display state, such as the Republic of China Invention No. 92116702 "Contrast Method for Contrast Enhancement (CONTRAST ENHANCEMENT)", No. 92116703 "Contrast Enhancement Using Sliding Window Method" (CONTRAST ENHANCEMENT) Method, No. 93107846 "Image Contrast Adjustment Method", No. 93113506 "Technology for Correcting the Appearance of Uneven Images Using Small-Scale Data", No. 94,712, 524 "Image Adjustment Method and Apparatus" and No. 94327726" The image edge enhancement device and its method are shown in the patent case.

雖然直方圖等化法近年來已被廣泛的運用於影像品質之改善，但仍然有許多可以改善的地方，例如查表式直方圖等化法，如第94132726號專利，需要將龐大的表格資訊載入記憶體中，方能進行轉換匹配動作，且若要有良好的補償效果，表格所紀錄的點數不能太少，但當表格紀錄點數增加時，意謂著需耗費更大量的記憶體空間與進行查表轉換的處理時間、影響處理器運算速度，並須較大之硬體代價，顯然仍有改善之處。Although histogram equalization has been widely used in imaging in recent years. Improvements in quality, but there are still many areas for improvement, such as the look-up table histogram equalization method, such as the patent No. 94327726, which requires huge form information to be loaded into the memory in order to perform the conversion matching action, and if To have a good compensation effect, the number of points recorded in the form should not be too small, but when the number of record points in the table increases, it means that it takes a lot of memory space and processing time for table lookup conversion, affecting processor operations. Speed, and a large hardware cost, obviously there is still improvement.

本發明之主要目的即在提供一種影像品質強化之方法，其不僅可確實強化影像之對比品質，並僅需較小之硬體代價、可縮短硬體之運算時間而提升運算速度，俾有效提升監視系統之即時影像品質及強化監視系統之監視效果。The main object of the present invention is to provide a method for enhancing the image quality, which not only can enhance the contrast quality of the image, but also requires a small hardware cost, can shorten the operation time of the hardware, and improve the operation speed, thereby effectively improving Monitor the real-time image quality of the system and monitor the monitoring system.

緣是，為達成前述之目的，本發明係提供一種影像品質強化之方法，其步驟至少包含有：擷取影像：擷取欲強化影像品質之一原始影像；抽樣(sampling)：以預定之抽樣率自該原始影像之像素抽樣而產生一抽樣影像；量化(quantization)：對該抽樣影像進行量化處理而產生一抽樣並量化之影像；製作查詢表：根據抽樣並量化之影像製作一查詢表(look－up table)；轉換獲得新影像：透過該查詢表將原始影像轉換而獲得一影像品質強化之新影像。Therefore, in order to achieve the foregoing object, the present invention provides a method for enhancing image quality, the method comprising at least: capturing an image: extracting an original image to enhance image quality; sampling: sampling by predetermined Generating a sample image from the pixel of the original image; quantification: quantizing the sampled image to produce a sampled and quantized image; creating a lookup table: creating a lookup table based on the sampled and quantized image ( Look-up table); Convert to obtain a new image: Convert the original image through the lookup table to obtain a new image with enhanced image quality.

此外，本發明更提供一種影像品質強化之方法，其步驟至少包含有：擷取影像：擷取欲強化影像品質之一原始彩色影像；第一次色彩空間轉換：將原始彩色影像之色彩空間轉換到含亮度Y 的色彩空間；抽樣：以預定之抽樣率自該原始彩色影像之亮度Y 抽樣而產生一抽樣亮度；量化：對該抽樣亮度進行量化處理而產生一抽樣並量化之亮度；製作查詢表：根據抽樣並量化之亮度製作一查詢表；轉換獲得新亮度：透過該查詢表將原始彩色影像亮度Y 轉換而獲得一品質強化彩色影像之新亮度；第二次色彩空間轉換：將含新亮度之強化彩色影像之色彩空間轉換回原來之色彩空間。In addition, the present invention further provides a method for enhancing image quality, the steps of which include at least: capturing an image: extracting an original color image that is to enhance image quality; and first color space conversion: converting a color space of the original color image To the color space containing the brightness Y ; sampling: sampling a brightness from the brightness Y of the original color image at a predetermined sampling rate; quantification: quantizing the sampled brightness to produce a sampled and quantized brightness; Table: Create a lookup table based on the sampled and quantized brightness; convert to obtain new brightness: convert the original color image brightness Y through the lookup table to obtain a new brightness of a quality enhanced color image; second color space conversion: will contain new The color space of the enhanced color image of the brightness is converted back to the original color space.

以下，茲舉本發明數個較佳實施例，並配合圖式做進一步之詳細說明如後：首先，請參閱圖一所示，本發明一較佳實施例之影像品質強化之方法，可用以強化灰階影像之品質，其第一步驟係擷取影像100：先擷取欲強化影像品質之一原始影像f ，該原始影像f 係灰階影像。In the following, a number of preferred embodiments of the present invention will be described in detail with reference to the drawings. First, referring to FIG. 1 , a method for image quality enhancement according to a preferred embodiment of the present invention can be used. strengthen the quality of the gray scale image, a first captured image 100 based step: capturing the first one to strengthen the image quality of the original image f, f Represents the original image gray scale image.

本發明之第二步驟係定義參數110：係定義該原始影像f 之像素尺寸、列、行抽樣率及灰階量化之尺寸值等參數，該原始影像f 之像素尺寸參數係M ×N 像素，列抽樣率之參數係1/p 、行抽樣率之參數係1/q ，灰階量化之尺寸參數QL 。The second step of the present invention based defined parameter 110: line defines the pixel size of the original image f, the column, row sampling rate and size of the gray scale of the quantization parameters, the original image pixel size f of M × N pixels based parameters, The parameters of the column sampling rate are 1/ p , the parameter of the sampling rate of the line is 1/ q , and the size parameter QL of the grayscale quantization.

本發明之第三步驟係抽樣120：係以列抽樣率1/p 、行抽樣率1/q 自原始影像f 中抽樣而產生一抽樣影像f _S ，該抽樣影像f _S 之像素尺寸係M _S ×N _S 像素，旦，M _S = M /p 為小於等於M /p 之最大整數、N _S = N /q 為小於等於N /q 之最大整數。The third step of the present invention is sampling 120: sampling a sample image f _S from the original image f at a column sampling rate of 1/ p and a line sampling rate of 1/ q , and the pixel size of the sampled image f _S is M _S × N _S pixels, Dan, M _S = M / p Is the largest integer less than or equal to M / p , N _S = N / q Is the largest integer less than or equal to N / q .

本發明之第四步驟係量化130：係基於灰階量化之尺寸參數QL 將該抽樣影像f _S 量化處理為另一抽樣並量化之影像f _SQ ，該影像f _SQ 在位置(x ,y )之灰階值為f _SQ (x ,y )，且，f _SQ (x ,y )= f _S (x ,y )/QL 為小於等於f _S (x ,y )/QL 之最大整數。The fourth step of the present invention quantizes 130: quantizing the sampled image f _S into another sampled and quantized image f _SQ based on the grayscale quantization size parameter QL , the image f _{SQ being} at position ( x , y ) The gray scale value is f _SQ ( x , y ), and f _SQ ( x , y )= f _S ( x , y ) / QL Is the largest integer less than or equal to f _S ( x , y ) / QL .

本發明之第五步驟係製作查詢表140：係依據該抽樣並量化之影像f _SQ 製作查詢表T ，假設在f _SQ 中，灰階值為k 的像素有n _k 個，則T 的定義如下：其中round 表取四捨五入，i =0,1,L,255/QL ，j =QL ×i ,QL ×i +1,L,min{QL ×(i +1)-1,255}。The fifth step of the present invention is to create a lookup table 140: based on the sampled and quantized image f _{SQ to} make a lookup table T , assuming that in f _SQ , there are n _k pixels with a grayscale value of k , then T is defined as follows : The round table is rounded off, i =0, 1, L, 255/ QL , j = QL × i , QL × i +1, L, min{ QL ×( i +1)-1,255}.

本發明之最後步驟係轉換獲得新影像150：係透過該查詢表T 將原始影像f 轉換為一新影像f _e ，其方法如下：f _e (x ,y )=T (f (x ,y )) 其中0 x M -1且0 y N -1。The final step of the present invention is to convert a new image 150: the original image f is converted into a new image f _e through the lookup table T , as follows: f _e ( x , y ) = T ( f ( x , y ) Where 0 x M -1 and 0 y N -1.

藉此，即可獲致對比品質強化之灰階影像。In this way, a grayscale image with contrast quality enhancement can be obtained.

以下，茲舉數例並利用MATLAB軟體模擬運算說明本發明用以強化灰階影像品質之特色與效果，MATLAB軟體係習知之數學運算軟體，此處不另贅述。In the following, the MATLAB software simulation operation is used to illustrate the characteristics and effects of the present invention for enhancing the quality of grayscale images. The mathematical operation software of the MATLAB software system is not described here.

如表一至表三及附件一、附件二所示，係依本發明方法將灰階量化尺寸參數QL 分別設為2、3、4(QL =2、3、4)時，在各列與行的抽樣率下所獲得品質強化之灰階影像與利用習知直方圖等化法(即未抽樣且未量化(p =q =1，QL =1))處理所獲得灰階影像之比較，表一至表三包含平均每個像素絕對值差的期望值(μ)與標準差(σ)： As shown in Tables 1 to 3 and Annexes I and II, the gray scale quantization size parameter QL is set to 2, 3, 4 ( QL = 2, 3, 4) according to the method of the present invention, in each column and row. Comparison of the gray-scale image obtained by the quality of the sample at the sampling rate and the gray-scale image obtained by the conventional histogram equalization method (ie, unsampled and unquantized ( p = q =1, QL =1)) From Table 1 to Table 3, the expected value (μ) and standard deviation (σ) of the absolute difference between each pixel are averaged:

其次，如表四及附件三所示，若僅抽樣而未進行量化(灰階量化尺寸參數QL =1)時，灰階影像在各列與行的抽樣率下所獲得結果與利用習知直方圖等化法(即未抽樣且未量化(p =q =1，QL =1))處理所獲得結果之比較，表四包含平均每個像素絕對值差的期望值(μ)與標準差(σ)： Secondly, as shown in Table 4 and Annex III, if only sampling is performed without quantization (gray scale quantization size parameter QL =1), the gray-scale image is obtained at the sampling rate of each column and row and the conventional histogram is used. Comparison of the results obtained by the graph equalization method (ie unsampled and unquantized ( p = q =1, QL =1)), Table 4 contains the expected value (μ) and standard deviation (σ) of the average absolute difference of each pixel. ):

由上可知，本發明之方法確實可強化灰階影像之對比品質，且，選用Altera公司編號EP1S80F1508I7之stratix系列FPGA晶片；經由QuartusII積體電路設計硬體驗證軟體驗證證明，應用本發明方法強化灰階影像之品質，其三級管線化超大型積體電路的設計，其產出率、硬體代價皆具有明顯較習知直方圖等化法為佳之效果。It can be seen from the above that the method of the present invention can strengthen the contrast quality of the gray-scale image, and the stratix series FPGA chip of Altera Company No. EP1S80F1508I7 is selected; the hardware verification software verification by the Quartus II integrated circuit design proves that the method of the present invention is used to strengthen the gray The quality of the order image, the design of its three-stage pipelined ultra-large integrated circuit, its output The rate and hardware cost are obviously better than the conventional histogram equalization method.

如表五所示，顯示在習知直方圖等化法(即未抽樣且未量化(p =q =1，QL =1))的三級管線化超大型積體電路設計之產出率與硬體代價： As shown in Table 5, the output rate of the three-stage pipelined ultra-large integrated circuit design shown in the conventional histogram equalization method (ie, unsampled and unquantized ( p = q =1, QL =1)) Hardware cost:

如表六所示，顯示在抽樣但未量化(p =q =2,QL =1)時的三級管線化超大型積體電路設計的產出率與硬體代價： As shown in Table 6, the output rate and hardware cost of a three-stage pipelined ultra-large integrated circuit design when sampling but not quantified ( p = q = 2, QL =1):

如表七所示，顯示本發明在抽樣並量化(p =q =2,QL =2)時的三級管線化超大型積體電路設計的產出率與硬體代價： As shown in Table 7, the yield and hardware cost of the three-stage pipelined ultra-large integrated circuit design of the present invention when sampling and quantizing ( p = q = 2, QL = 2) are shown:

由表五及表七可知，本發明在p =q =2與QL =2時，其三級管線化超大型積體電路設計的產出率為4018.0 fps(畫面/秒)，遠大於習知直方圖等化法之三級管線化超大型積體電路設計的產出率876.36 fps，且，此時本發明的硬體代價為14857 LEs，遠低於習知直方圖等化法的硬體代價26098 Les，顯然本發明可獲致較高之產出率與較低之硬體代價。It can be seen from Tables 5 and 7 that when p = q = 2 and QL = 2, the yield of the three-stage pipelined ultra-large integrated circuit design is 4018.0 fps (pictures per second), which is much larger than the conventional one. The yield of the three-stage pipelined ultra-large integrated circuit design of the histogram equalization method is 876.36 fps, and at this time, the hardware cost of the invention is 14857 LEs, which is much lower than that of the conventional histogram equalization method. At the cost of 26098 Les, it is clear that the present invention achieves a higher yield and a lower hardware cost.

基此，本發明不僅可縮短硬體之運算時間、提升運算速度，更可獲致較低之硬體代價，顯然較需較大硬體代價及硬體運算時間之習知直方圖等化法更適用於監視系統之即時影像品質提升，而具備更佳之實用價值。Therefore, the present invention not only can shorten the operation time of the hardware, improve the operation speed, but also obtain a lower hardware cost, and obviously requires a larger hardware cost and a conventional histogram equalization method of the hardware operation time. It is suitable for monitoring the real-time image quality of the system, and has better practical value.

如圖二所示，係本發明另一較佳實施例之影像品質強化之方法，係可用以強化彩色影像之品質，其步驟如下：As shown in FIG. 2, the image quality enhancement method according to another preferred embodiment of the present invention can be used to enhance the quality of a color image, and the steps are as follows:

第一步驟係擷取影像200：擷取欲強化影像品質之一原始彩色影像。The first step is to capture the image 200: extracting one of the original color images of the image quality to be enhanced.

第二步驟係定義參數210：係定義該原始彩色影像之像素尺寸、列、行抽樣率及亮度量化之尺寸值等參數，該原始彩色影像之像素尺寸參數係M ×N 像素，列抽樣率之參數係1/p 、行抽樣率之參數係1/q ，亮度量化之尺寸參數QL 。The second step defines a parameter 210: defining a pixel size, a column, a row sampling rate, and a size value of the brightness quantization of the original color image. The pixel size parameter of the original color image is M × N pixels, and the column sampling rate is The parameter is 1/ p , the parameter of the line sampling rate is 1/ q , and the size parameter QL of the brightness quantization.

第三步驟係第一次色彩空間轉換220：係將原始彩色影像之色彩空間轉換到含亮度Y 的色彩空間，獲致一原始亮度Y 。The third step is the first color space conversion 220: converting the color space of the original color image to the color space containing the brightness Y to obtain an original brightness Y.

第四步驟係抽樣230：係以列抽樣率1/p 、行抽樣率1/q ，自原始亮度Y 中抽樣而產生抽樣亮度Y _S ，該抽樣亮度Y _S 之像素尺寸係M _S ×N _S 像素，且，M _S = M /p 為小於等於M /p 的最大整數、N _S = N /q 為小於等於N /q 的最大整數。The fourth step is sampling 230: taking the sampling rate 1/ p and the line sampling rate 1/ q , sampling from the original brightness Y to generate the sampling brightness Y _S , the pixel size of the sampling brightness Y _S is M _S × N _S Pixel, and, M _S = M / p Is the largest integer less than or equal to M / p , N _S = N / q Is the largest integer less than or equal to N / q .

第五步驟係量化240：係基於亮度量化之尺寸參數QL 將該抽樣亮度Y _S 量化處理為另一抽樣並量化之亮度Y _SQ ，該Y _SQ 在位置(x ,y )之亮度值係Y _SQ (x ,y )，且，Y _SQ (x ,y )= Y _S (x ,y )/QL 為小於等於Y _S (x ,y )/QL 的最大整數。A fifth step based quantizer 240: Department quantization processing based on the size of the luminance quantization parameter QL Y _S The luminance samples and another sample of the quantized luminance Y _SQ, in which Y _SQ position (x, y) of the luminance value Y _SQ-based ( x , y ), and, Y _SQ ( x , y )= Y _S ( x , y ) / QL Is the largest integer less than or equal to Y _S ( x , y ) / QL .

第六步驟係製作查詢表250：係依據該抽樣並量化之亮度Y _SQ 製作查詢表T ，假設在Y _SQ 中，亮度為k 的像素有n _k 個，則T 的定義如下：其中round 表取四捨五入，i =0,1,L,255/Q L，j =QL ×i ,QL ×i +1,L,min{QL ×(i +1)-1,255}。A sixth step of making query-based table 250: lines according to the sampling and quantization of the luminance Y _SQ produced lookup table T, Y _SQ is assumed that the brightness of a pixel has n-k a _k, T are defined as follows: The round table is rounded off, i =0, 1, L, 255/ Q L , j = QL × i , QL × i +1, L, min{ QL ×( i +1)-1,255}.

第七步驟係轉換獲得新亮度260：係透過該查詢表T 將原始影像亮度Y 轉換為一新亮度Y _e ，方法如下：Y _e (x ,y )=T (Y (x ,y )),其中0 x M -1且0 y N -1。The seventh step is to convert the new brightness 260: the original image brightness Y is converted into a new brightness Y _e through the look-up table T , as follows: Y _e ( x , y )= T ( Y ( x , y )), Where 0 x M -1 and 0 y N -1.

本發明之最後步驟係第二次色彩空間轉換270：係將含新亮度之強化彩色影像之色彩空間轉換回原來之色彩空間。The final step of the present invention is a second color space conversion 270: converting the color space of the enhanced color image with the new brightness back to the original color space.

藉此，即可獲致品質強化之彩色影像。In this way, quality-enhanced color images can be obtained.

以下，茲舉數例並利用MATLAB軟體模擬運算說明本發明用以強化彩色影像品質之特色與效果：如表八至表十及附件四、附件五所示，係依本發明方法將亮度量化尺寸參數QL 分別設為2、3、4(QL =2、3、4)時，在各列與行的抽樣率下所獲得品質強化之彩色影像與利用習知直方圖等化法(即未抽樣且未量化(p =q =1，QL =1))處理所獲得彩色影像之比較結果，表八至表十包含平均每個像素絕對值差的期望值(μ)與標準差(σ)。In the following, the MATLAB software simulation operation is used to illustrate the characteristics and effects of the present invention for enhancing the quality of color images: as shown in Tables 8 to 10 and Annex IV and Annex V, the brightness is quantized according to the method of the present invention. When the parameter QL is set to 2, 3, and 4 ( QL = 2, 3, and 4, respectively), the color image of the quality enhancement obtained at the sampling rate of each column and row and the conventional histogram equalization method (ie, unsampling) And unquantized ( p = q =1, QL =1)) The result of the comparison of the obtained color images is processed. Tables 8 to 10 contain the expected value (μ) and standard deviation (σ) of the average absolute difference of each pixel.

其次，如表十一及附件六所示，若僅抽樣而未進行量化(亮度量化尺寸參數QL =1)時，擷取之彩色影像在各列與行的抽樣率下實施之結果與利用習知直方圖等化法(即未抽樣且未量化(p =q =1，QL =1))處理所獲得結果之比較，表十一包含平均每個像素絕對值差的期望值(μ)與標準差(σ)： Secondly, as shown in Table 11 and Annex 6, if only sampling is performed without quantization (luminance quantization size parameter QL =1), the results and utilization of the captured color image at each column and row sampling rate are used. Comparing the results obtained by the histogram equalization method (ie unsampled and unquantized ( p = q =1, QL =1)), Table 11 contains the expected value (μ) and the standard of the absolute difference between each pixel. Difference (σ):

基此，本發明之方法亦確實可強化彩色影像之對比品質。Accordingly, the method of the present invention also enhances the contrast quality of color images.

綜上所述，本發明所提供之影像品質強化之方法，其透過抽樣及量化之步驟，除了可確實強化灰階及彩色影像之對比品質之外，相較於習知直方圖等化法，更可縮小硬體代價、縮短硬體運算時間、提升運算速度，俾可提升監視系統之即時影像品質及強化監視系統之監視效果；緣是，本發明確實符合發明專利之要件，爰依法提出申請。In summary, the image quality enhancement method provided by the present invention, through the steps of sampling and quantifying, can not only strengthen the contrast quality of gray scale and color image, but also compares with the conventional histogram equalization method. It can reduce the hardware cost, shorten the hardware computing time, and improve the computing speed. It can improve the real-time image quality of the monitoring system and strengthen the monitoring effect of the monitoring system. The reason is that the invention does meet the requirements of the invention patent and submits an application according to law. .

擷取影像‧‧‧100Capture images ‧‧100

定義參數‧‧‧110Define parameters ‧‧‧110

抽樣‧‧‧120Sampling ‧ ‧ ‧

量化‧‧‧130Quantification ‧‧‧130

製作查詢表‧‧‧140Making a questionnaire ‧‧140

轉換獲得新影像‧‧‧150Convert to get new images ‧‧‧150

擷取影像‧‧‧200Capture images ‧‧200

定義參數‧‧‧210Define parameters ‧‧‧210

第一次色彩空間轉換‧‧‧220First color space conversion ‧‧‧220

抽樣‧‧‧230Sampling ‧ ‧ 230

量化‧‧‧240Quantification ‧‧‧240

製作查詢表‧‧‧250Making a questionnaire ‧ ‧ 250

轉換獲得新亮度‧‧‧260Conversion gains new brightness ‧‧‧260

第二次色彩空間轉換‧‧‧270Second color space conversion ‧‧‧270

圖一係本發明一較佳實施例之流程圖。1 is a flow chart of a preferred embodiment of the present invention.

圖二係本發明另一較佳實施例之流程圖。Figure 2 is a flow chart of another preferred embodiment of the present invention.

附件一係原始影像、習知直方圖等化法處理之灰階影像與QL =2時本發明在各列與行的抽樣率下所獲得品質強化之灰階影像。Annex 1 is a gray-scale image of the quality enhancement obtained by the present invention at the sampling rate of each column and row when the gray-scale image processed by the original image, the conventional histogram, and the like is QL =2.

附件二係原始影像、習知直方圖等化法處理之灰階影像與QL =4時本發明在各列與行的抽樣率下所獲得品質強化之灰階影像。Annex II of the original image based, conventional processing of gray scale image and the histogram of QL = 4, the method of the present invention is to strengthen the gray scale image quality at sampling rates of rows and columns obtained.

附件三係原始彩色影像、習知直方圖等化法處理之灰階影像與僅抽樣而未進行量化(QL =1)時在各列與行的抽樣率下所獲得之灰階影像。Annex 3 is the gray-scale image obtained by the original color image, the conventional histogram, etc., and the gray-scale image obtained at the sampling rate of each column and row when the sample is not sampled but not quantized ( QL =1).

附件四係原始彩色影像、習知直方圖等化法處理之彩色影像與QL =2時本發明在各列與行的抽樣率下所獲得品質強化之彩色影像。Annex 4 is a color image obtained by the original color image, the conventional histogram, etc., and the color image obtained by the present invention at the sampling rate of each column and row when QL = 2.

附件五係原始彩色影像、習知直方圖等化法處理之灰階影像與QL =4時本發明在各列與行的抽樣率下所獲得品質強化之彩色影像。Annex 5 is a color image of the quality enhancement obtained by the present invention at the sampling rate of each column and row when the gray scale image processed by the original color image, the conventional histogram, and the like is QL =4.

附件六係原始彩色影像、習知直方圖等化法處理之灰階影像與僅抽樣而未進行量化(QL =1)時在各列與行的抽樣率下所獲得之彩色影像。Annex 6 is a color image obtained at the sampling rate of each column and row when the grayscale image processed by the original color image, the conventional histogram, and the like is sampled without being quantized ( QL =1).

擷取影像‧‧‧100Capture images ‧‧100

定義參數‧‧‧110Define parameters ‧‧‧110

抽樣‧‧‧120Sampling ‧ ‧ ‧

量化‧‧‧130Quantification ‧‧‧130

製作查詢表‧‧‧140Making a questionnaire ‧‧140

轉換獲得新影像‧‧‧150Convert to get new images ‧‧‧150

Claims

一種影像品質強化之方法，其步驟至少包含有：擷取影像：擷取欲強化影像品質之一原始影像f ，該原始影像f 係灰階影像，其像素尺寸參數係M ×N 像素；定義參數：係定義該原始影像之像素尺寸、列、行抽樣率及灰階量化之尺寸值等參數；抽樣：以預定之抽樣率自該原始影像之像素抽樣而產生一抽樣影像f _S ，該抽樣影像f _S 係以列抽樣率1/p 、行抽樣率1/q 自原始影像f 中抽樣而產生，該抽樣影像f _S 之像素尺寸係M _S ×N _S 像素，且，M _S = M /p 為小於等於M /p 之最大整數、N _S = N /q 為小於等於N /q 之最大整數；量化：對該抽樣影像進行量化處理而產生一抽樣並量化之影像，係基於灰階量化之尺寸參數QL 將該抽樣影像f _S 量化處理為另一抽樣並量化之影像f _SQ ，該影像f _SQ 在位置(x ,y )之灰階值為f _SQ (x ,y )，且，f _SQ (x ,y )= f _S (x ,y )/QL 為小於等於f _S (x ,y )/QL 之最大整數；製作查詢表：根據抽樣並量化之影像製作一查詢表；及轉換獲得新影像：透過該查詢表將原始影像轉換而獲得一影像品質強化之新影像。A method for enhancing image quality, the method comprising at least: capturing an image: capturing an original image f of an image quality to be enhanced, the original image f being a grayscale image, the pixel size parameter being M × N pixels; defining parameters : defining a pixel size, a column, a row sampling rate, and a grayscale quantization size value of the original image; sampling: sampling a pixel from the original image at a predetermined sampling rate to generate a sample image f _S , the sample image The f _S is generated by sampling from the original image f at a column sampling rate of 1/ p and a line sampling rate of 1/ q . The pixel size of the sampled image f _S is M _S × N _S pixels, and M _S = M / p Is the largest integer less than or equal to M / p , N _S = N / q Is the largest integer less than or equal to N / q ; quantization: quantizing the sampled image to generate a sampled and quantized image, and quantizing the sampled image f _S into another sample based on the gray-scale quantization size parameter QL The quantized image f _SQ , the gray level value of the image f _SQ at the position ( x , y ) is f _SQ ( x , y ), and f _SQ ( x , y )= f _S ( x , y ) / QL Is the largest integer less than or equal to f _S ( x , y ) / QL ; making a lookup table: creating a lookup table based on the sampled and quantized image; and converting to obtain a new image: converting the original image through the lookup table to obtain an image quality Enhanced new images.

如申請專利範圍第1項所述影像品質強化之方法，其中，製作查詢表之步驟中，係依據該抽樣並量化之影像f _SQ 製作查詢表T ，假設在f _SQ 中，灰階值為k 的像素有n _k 個，則T 的定義如下：其中round 表取四捨五入，i =0,1,L,255/QL ，j =QL ×i ,QL ×i +1,L,min{QL ×(i +1)-1,255}。The patentable scope of application of the method to strengthen the image quality to item 1, wherein the step of forming a lookup table, the system according to the sampling and quantization of the image lookup table T f _SQ produced, assuming f _SQ, the gray scale value k The pixels have n _k , then T is defined as follows: The round table is rounded off, i =0, 1, L, 255/ QL , j = QL × i , QL × i +1, L, min{ QL ×( i +1)-1,255}.

如申請專利範圍第2項所述影像品質強化之方法，其中，轉換獲得新影像之步驟中，透過該查詢表T 將原始影像f 轉換為一新影像f _e 之方法如下：f _e (x ,y )=T (f (x ,y ))其中0 x M -1且0 y N -1。The method of image quality enhancement according to claim 2, wherein in the step of converting the obtained new image, the method of converting the original image f into a new image f _e through the lookup table T is as follows: f _e ( x , y )= T ( f ( x , y )) where 0 x M -1 and 0 y N -1.

一種影像品質強化之方法，其步驟至少包含有：擷取影像：擷取欲強化影像品質之一原始彩色影像，其像素尺寸參數係M ×N 像素；定義參數：係定義原始彩色影像之像素尺寸、列、行抽樣率及亮度量化之尺寸值等參數；第一次色彩空間轉換：將原始彩色影像之色彩空間轉換到含亮度Y 的色彩空間；抽樣：以預定之抽樣率自該原始彩色影像之亮度Y 抽樣而產生一抽樣亮度Y _S ，該抽樣亮度Y _S 係以列抽樣率1/p 、行抽樣率1/q 自原始亮度Y 中抽樣而產生，該抽樣亮度Y _S 之像素尺寸係M _S ×N _S 像素，且，M _S = M /p 為小於等於M /p 的最大整數、N _S = N /q 為小於等於N /q 的最大整數；量化：對該抽樣亮度進行量化處理而產生一抽樣並量化之亮度；製作查詢表：根據抽樣並量化之亮度製作一查詢表；轉換獲得新亮度：透過該查詢表將原始彩色影像亮度Y 轉換而獲得一品質強化彩色影像之新亮度；及第二次色彩空間轉換：將含新亮度之強化彩色影像之色彩空間轉換回原來之色彩空間。A method for enhancing image quality, the method comprises at least: capturing an image: capturing an original color image of one of the image qualities to be enhanced, the pixel size parameter is M × N pixels; defining parameters: defining a pixel size of the original color image Parameters such as column, row sampling rate and brightness quantization size; first color space conversion: converting the color space of the original color image to the color space containing the brightness Y ; sampling: from the original color image at a predetermined sampling rate the luminance Y samples to generate a sampled luminance Y is _S, the sampling luminance Y _S line in a column sampling rate 1 / p, line sample rate of 1 / q from the original luminance Y of the samples produced, the pixels of the sampled luminance Y _S of dimension lines M _S × N _S pixels, and, M _S = M / p Is the largest integer less than or equal to M / p , N _S = N / q Is the largest integer less than or equal to N / q ; quantization: quantize the sampled brightness to produce a sampled and quantized brightness; create a lookup table: create a lookup table based on the brightness of the sampled and quantized; convert to obtain a new brightness: through the The lookup table converts the original color image brightness Y to obtain a new brightness of the quality enhanced color image; and the second color space conversion: converts the color space of the enhanced color image with the new brightness back to the original color space.

如申請專利範圍第4項所述影像品質強化之方法，其中，量化之步驟中，係基於亮度量化之尺寸參數QL 將該抽樣亮度Y _S 量化處理為另一抽樣並量化之亮度Y _SQ ，該Y _SQ 在位置(x ,y )之亮度值係Y _SQ (x ,y )，且，Y _SQ (x ,y )= Y _S (x ,y )/QL 為小於等於Y _S (x ,y )/QL 的最大整數。The image quality enhancement method according to claim 4, wherein in the step of quantizing, the sampled brightness Y _{S is} quantized into another sampled and quantized brightness Y _SQ based on the brightness parameter size parameter QL . The luminance value of Y _SQ at position ( x , y ) is Y _SQ ( x , y ), and Y _SQ ( x , y )= Y _S ( x , y ) / QL Is the largest integer less than or equal to Y _S ( x , y ) / QL .

如申請專利範圍第5項所述影像品質強化之方法，其中，製作查詢表之步驟中，係依據該抽樣並量化之亮度Y _SQ 製作查詢表T ，假設在Y _SQ 中，亮度為k 的像素有n _k 個，則T 的定義如下：其中round 表取四捨五入，i =0,1,L,255/QL ，j =QL ×i ,QL ×i +1,L,min{QL ×(i +1)-1,255}。The patentable scope of application of the method to strengthen the image quality to item 5, wherein the step of forming a lookup table, the system according to the sampling and quantization of the luminance Y _SQ produced lookup table T, Y _SQ is assumed that the brightness of a pixel k There are n _k , then T is defined as follows: The round table is rounded off, i =0, 1, L, 255/ QL , j = QL × i , QL × i +1, L, min{ QL ×( i +1)-1,255}.

如申請專利範圍第6項所述影像品質強化之方法，其中，轉換獲得新亮度之步驟中，透過該查詢表T 將原始彩色影像亮度Y 轉換為一新亮度Y _e 之方法如下：Y _e (x ,y )=T (Y (x ,y ))，其中0 x M -1且0 y N -1。The method for enhancing image quality according to claim 6 of the patent application, wherein, in the step of converting the obtained new brightness, the method of converting the original color image brightness Y into a new brightness Y _e through the look-up table T is as follows: Y _e ( x , y )= T ( Y ( x , y )), where 0 x M -1 and 0 y N -1.