TWI235007B - A method of image processing of low illumination and image processor - Google Patents

Abstract

A method of image processing of low illumination is adapted to a CMOS image sensor. When the average brightness of a image is less than the noise threshold limit value of the CMOS image sensor, the noise of the CMOS image sensor can be deleted by the method of the middle-value filtering calculation. The advantage is that the processing of calculation is not too complicated to occupy too much resource of a system so that it is adapted to real-time processing to present a good quality. Therefore, CMOS sensor can present a good quality whenever the image is in a state of high illumination or low illumination.

Description

Λ: ._, ίί—Zid No. 92130891_ year, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, week, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month, month of the year . In the past, given the booming communications industry and digital technology, personal digital assistants (PDAs), handheld computers (Pocket PCs), portable computers (Portable PCs), and mobile phones (Mobile Phones) were paired with digital cameras. (Digital Camera) is already a trend. Generally speaking, image sensors can be mainly divided into two types: photo-coupled devices (CCD) and complementary metal oxide semi-conductors (CMOS). Both use the same photoelectric conversion process to convert image incident light into voltage. Based on the consideration of power saving and low cost, the use of photo-charge coupled devices (CCD) is expensive due to the special process, and the driving circuit must operate at high voltage to make the power consumption high. In contrast, CMOS image sensors are What are the advantages of power saving and easy integration with peripheral components? The price is affordable, and it has gradually received attention in the low-end imaging product application market. In terms of product applications, CMOS image sensors are currently mainly used for computer cameras (PC Cameras) and low-order digital cameras. They can also be used in mobile phones, PDAs, video phones, fingerprint readers and other products. Although the CM0S image sensor has integrated system characteristics, it can be easily integrated with other control circuits, analog digital circuits (a / d Converter), or digital signal processing (Dsp) circuits on the same chip to achieve the so-called Target of the System On a Chip (SOC)

-.. 丄:-、% -7 lai_

f V. Description of the invention (2) There are still several problems to be overcome with CMOS image sensors. One is that in the process of integrating all components, the read noise increases due to the current amplification, which reduces the image quality. The second is that after integrating multiple functions, a new circuit is bound to be added. This will make the single pixel's liiumination area on the CM0s image sensor only the CCD image sensor. About 30%, resulting in poor light quality in dim light. . Figure 1 / A square camera zmi1 ® that is not familiar with the image processing flow of a digital camera. After the image 100 is focused through the lens 110, it is cast on a CM sensor 120, and the captured image is passed through The number of Γ of the posterior = Λ processor (D sp) 13 0 and a microprocessor (c ρ ϋ) 14. The image quality is better, but because the calculation process of signal processor 0 is more Complex enough to take up too much system resources. This method is not used on a PDA. Instead, a CMOS image sensor is used, and the image taken is converted by an analog / digital converter (ADC) to convert a class 1 signal into a digital signal for microprocessing. (Cpu) for imaging, so the quality of the daytime surface is affected. Therefore, although the 'CM0s' image sensor will have good image quality under normal illumination, the quality effect at low illumination is still not good. SUMMARY OF THE INVENTION “The object of the present invention is to provide a low-illuminance image, and the method is“ applicable to a CMOS image sensor to remove noise generated by low-illuminance ”without occupying too much system resources. . In order to achieve the above object of the present invention, the present invention proposes a low-light image processing method, which is suitable for processing one of the images captured by a CMOS image sensor.

12073twf1.ptc Page 7 a-flame — a 1 丨 $ # 92130891_ Car Moon Day Correction V. Description of the invention (3) Image 'This image has one of these pixels in most of the calculations in the calculation system. CMOS image sensing Noise filter processing. In order to achieve the above object of the present invention, the flash memory of a CMOS image sensor median filtering algorithm is mainly used to capture an image. This image package is lower than a noise of the CMOS image sensor and generates a noise in the image. on. In the application of the value-value filtering algorithm, when the processor is connected to the average brightness value of one of these pixels in the CMOS image sensor, the input image to median filtering according to a preferred embodiment of the present invention further includes a buffer configured in After reaching the buffer, it is output. Because the invention adopts the noise of median filtering, its advantage is the system resources of the calculation process, which is suitable for real-time processing of CMOS image sensors regardless of high-quality performance. In order to make the above of the present invention and its comprehensibility easier, a preferable feature is enumerated below. This low-illumination image processing method has an average brightness value. When the average brightness value is small, a median value is performed on the image. The present invention proposes an image processing device, a microprocessor, and a storage body. CMOS image sensor, including many pixels, when the brightness of the image is critical, the CMOS image sensor and flash memory system are used to store one for a median filtering process, and micro and flash memory Volume, used to calculate the image, when the average brightness value is less than the noise threshold algorithm in the application program. For example, at the rear of the above-mentioned image processing equipment processor, the image input algorithm to remove the CMOS image sensor will not be so complicated that it will take up too much to present the processing quality. Therefore, the degree or low illumination will have quite good. Other purposes, features, and advantages can be more clear. With the accompanying drawings, details

12073twf1.ptc Page 8 θ ........ 92130891_t Yue Yuezheng Zheng __ V. Description of the invention (4) The detailed description is as follows: For implementation, please refer to page FIG. 2 is a flow block diagram showing a low-illuminance image processing device according to a preferred embodiment of the present invention, which is applicable to a C M 0 S image sensor. Under normal illumination, such as daytime photography, or capturing images in well-lit places such as outdoor suburbs, the CMOS image sensor 210 has good image quality, but in low illumination, such as night photography, or indoor lighting In the case of dim light, such as insufficient lighting, the effect of the image is not good, and noise may even be generated, affecting the quality of the image. Since the C M 0 S image sensor 2 10 only appears noise at low illumination, the present invention calculates an image of 200, and calculates the average brightness value to determine that low illumination is performed in the case of insufficient light. Image processing. Please refer to FIG. 2. After a CMOS image sensor 21 captures an image 200, a microprocessor 2 2 0 can be used to calculate the average brightness value of each pixel of the image 2 2. Under dim conditions, the average brightness value of the image is significantly lower. At this time, the CMOS image sensor 2 1 0 increases the chance of generating noise. Since the degree of the image 2 0 directly affects the noise generated by the c M 0 S image sensor 2 1 0, the average brightness value can represent the best light intensity accepted by the pixel block, but it can also be in other ways Calculate the brightness accepted by the pixel block, such as the average brightness value or the best brightness value obtained by capturing the daytime pixel block in the central area. In addition, when the average brightness value is less than one of the noise thresholds of the CMOS image sensor 21, it means that the image 2000 is dark, but the actual image quality is poor, and the image 200 will appear noise. In this embodiment, when [μ 〇 s

12073twf1.ptc Page 9

92130891 V. Description of the invention (5)

When the image sensor 2 10 generates noise, the application of a median wave algorithm can be used to remove the noise on the image. This threshold of noise " can be self-defeating on the basis of α roundness according to the actual situation of the user. Taking 8-bit image processing as an example, the brightness of image 200 is divided from small to large into 256 calculation units (range 0 ~ 2 5), and the noise threshold can be set to 60 or other experience value. Then, the Y value, the Y value, and the V value of the image in the book element block are lifted (Y represents brightness, u ^ Table I "tone, and V represents saturation) or corresponding R, G, and B values (R ^ Table red light, G stands for green light, B stands for I light) 'where the day element block is made up of n rows and n columns of book element arrays, and N is an integer greater than 1, and enter these light output & Application of chroma value and saturation to a median filtering algorithm g ^ to remove noise on the image 2000. The median filtered scene > image processing technology is an image processing technology to remove noise '' The main spirit of its algorithm is to take appropriate pixel blocks in an image, such as 3x3 pixel blocks or 6x6 pixel blocks to take the median value. When noise occurs in a pixel block, Then the yuv value (γ stands for brightness, u stands for hue, v stands for saturation) or RGB values ^ (R stands for red light, G stands for green light, and B stands for blue light) are bound to _ appear numerically, using the median Instead of abnormal data, you can remove noise if the value is abnormal. The advantage of this algorithm is that it has been calculated The process will not be too complicated to occupy too much system resources, and is suitable for real-time (real time processing) to present the quality of image processing. «Refer to Figures 2 and 3 for the month of month, where Figure 3 shows a low level of the present invention Flow chart of image processing equipment and method for illuminance. In step 丨, because CMOS sensor 21 〇 noise will only appear in the case of low illuminance

II-No. 92130891 Birth month B_ Correction V. Description of the invention (6) The message 'so you can first calculate the average brightness value of the image 2 0 0, and then in step 2' the scene > the average brightness value of the image 2 0 0 (Gamma value is less than 60 or other empirical values) When the noise value is less than the threshold value, the image is processed and median filtering is performed to remove noise (such as step 3), and when the brightness is sufficient, there is no need to do any Process (as in step 4). In this way, more system resources can be saved, and the CM0s image sensor 2 10 has a very good quality performance in both high and low light. In addition, in step 3, when it is detected that the brightness is insufficient, the present invention can also adjust the CMOS image sensor 2 1 0 to the maximum sensitivity, so the processed image 2 0 only needs to pass through The application of the value filtering algorithm 2 3 0 will have good image quality. In addition, if the sensitivity of the CMOS image sensor 210 itself is not set at the maximum value, in order to enhance the brightness of the image, an image equalizer or other image processing method must be used, although the median filtering algorithm is then used 2 3 0 The application to remove noise will also not improve much because the image quality of c Μ 0 S image sensor 2 1 0 is too poor. In addition, in the second figure, a PDA and a mobile phone, as well as a portable computer, each have a built-in microprocessor 2 2 0 and a memory device 2 4 0, such as a flash memory (F 1 ash M em 〇 ry), so there is no need to use the digital signal processor (DSP) of the digital camera, but an application program that uses the median filtering algorithm stored in the external flash memory to remove noise and save system resources. In addition, for real-time development (rea 1 time), whether or not the image processed image 2 0 0 should be developed after a buffer 2 5 0, will not let the user There is a delay in the calculation processing time during high-low brightness conversion.

Page 11—year, month, and green number 92130891_year, month, day and year__ V. Description of the invention (7) In summary, the low-illumination image processing method of the present invention is based on the average brightness value of the image being less than the CMOS image sensor When the noise threshold of the sensor is used, the application of the median filtering algorithm is used to remove the noise of the CMOS image sensor. The advantage is that the calculation process will not be too complicated to occupy too much system resources, which is suitable for real-time processing. To present the quality of processing. Therefore, the C M 0 S image sensor will perform fairly well in both high and low light conditions. Although the present invention has been disclosed as above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some changes and retouch without departing from the spirit and scope of the present invention. The scope of protection of the invention shall be determined by the scope of the attached patent application.

12073twf1.ptc Page 12 mwmt 1¾ No. 92130891 Revised Brief Description of Drawings Figure 1 is a block diagram showing the image processing flow of a conventional digital camera. FIG. 2 is a flow block diagram of a low-luminance image processing device according to a preferred embodiment of the present invention. FIG. 3 is a flowchart of a low-illuminance image processing device and a method thereof according to the present invention. Schematic icon 100 110 120 130 140 200 210 220 230 240 250 Description] Image Lens CMOS Image Sensor Digital Signal Processor Microprocessor Image CMOS Image Sensor Microprocessor Median Filter Algorithm Memory Device Buffer

12073twf1.ptc Page 13

Claims (1)

Year month ________ No. 92130891_ year month day amendment_ VI. Patent application scope 1. A low-light image processing method for processing an image captured by a CMOS image sensor, the image includes a plurality of images The low-illumination image processing method includes the following steps: calculating an average brightness value of one of the pixels in the image; and when the average brightness value is less than a noise threshold, then: performing a median value on the image Filtering. 2. The low-illumination image processing method described in item 1 of the scope of patent application, wherein when the average brightness value is less than the noise threshold, the method further includes a step of increasing the sensitivity of the CMOS image sensor. 3. The low-illumination image processing method and method described in item 1 of the scope of patent application, further including inputting the image into a buffer and then outputting it. 4. An image processing device comprising: a CMOS image sensor for capturing an image including a plurality of pixels; a memory storing a median filter algorithm for storing a median filter algorithm An application program to perform a median filtering process; and a microprocessor connected to the CMOS image sensor and the flash memory to calculate an average brightness value of one of the daylight elements in the image, When the average brightness value is less than a noise threshold, the image is input to perform the median filtering process. 5. The image processing device as described in item 4 of the scope of patent application, further comprising a buffer disposed at the rear end of the microprocessor so that the image is input to the buffer and then output. 12073twf1.ptc Page 14