Common camera issues and solutions for kids cameras

Summary of problems and solutions for children’s camera lenses

1. Explanation of the name
White balance:
White balance refers to the sensor’s ability to accurately reproduce colors in environments with constantly changing lighting conditions. Most camera systems have an automatic white balance function, which automatically adjusts the white balance value as lighting conditions change. Design engineers are looking for image sensors equipped with a good automatic white balance (AWB) control to provide correct color reproduction. To talk about white balance, we must first establish the concept of color temperature. In fact, in the field of photography, light sources are mostly defined based on their color temperature. The unit of color temperature is Kelvin, and the color presented at different temperatures is color temperature. When a black object is heated, it begins to emit light, first turning dark red, then yellow, then white, and finally blue (you can observe the filament in a light bulb, but due to temperature limitations, you generally won’t see it turn blue). In short, this phenomenon is very common in daily life.

The human brain can carefully analyze the signals received from the eyes, thus perceiving different color temperatures to display the same white color. However, children’s camera lenses cannot do so. Photos taken in the morning tend to be reddish, while those taken at dusk tend to be yellowish. Even when the same white paper is photographed under different environments, such as at different times and with different light sources, there will be varying degrees of deviation. Adjusting white balance is to define white, so as to accurately record the colors we see with our eyes. The sensor provides two control methods for white balance: automatic white balance (AWB) and manual white balance (MWB). AWB is the default setting for the sensor’s internal ISP, which contains a complex rectangular diagram that determines the white balance reference point in the image, thus achieving white balance adjustment. Since children’s cameras do not belong to the category of cameras with high color requirements, we generally use AWB, which performs well when the light source is not particularly complex. MWB requires manually setting the gain values of R, G, and B, and there are dedicated registers for these three values.

Noise caused by incorrect edge selection in PCLK sampling:
After changing the polarity of Pclk, the photo taken of this impulsive point turns green. Perhaps the PCLK sampling edge is incorrect, you can try reversing PCLK. It could also be a problem with missing data cables. Example 3, as shown in the following figure. The issue was resolved by modifying the rising and falling edges of PCLK. There are two main points: 1. Modify the slope of the rising edge of PCLK. 2. Alternatively, modify the slope of the rising edge of the I/O
The reason is that the length of the wiring in the module layout of different manufacturers, the thickness of the FPC, which may affect the acquisition of PCLK, the tolerance of the FPC being too large, or whether there are any problems with the production of the head board, may all cause this problem. If the slope of the rising edge of PCLK can be changed through hardware, this problem can also be solved. The problem was solved by modifying the rising and falling edges of pclk

FPN issue:
There is no such problem during the day or in brighter areas, except when using a flash to take photos in low light conditions. FPN( fixed pattern noise), No solution.

Step effect:
If the gain is too large, multiplying the quantization step of digitization by a larger one will result in a step effect. It is also related to the insufficient internal quantization accuracy. In addition, if the gain of different color channels is different (the R/G/B gain calculated by white balance is different), a color phase error will occur. Only two channels, B and G, were drawn. If B is larger than G, it will cause grayscale scenery. In some places, B is larger than G, resulting in continuous color alternation.

Vertical stripes caused by power issues:
It has now been confirmed that it is a problem with the power supply. I have connected a large capacitor to each power supply, and the stripes have disappeared. Now I am using CPU I/O data collection, and the effect is very good.

Partial redness caused by mismatch between lens and camera:
The center of the lower part of the picture is slightly red. The engineer adjusted the LENS CORDION to its limit, but the problem still exists. It was confirmed that it was caused by a mismatch between LENS and SENSOR. After the module manufacturer replaced the lens, the problem was basically solved.
Firstly, your photo AWB is incorrect as it has not reached white balance
2. The jagged edges in the photo are very serious. 3. For color cast issues, you first need to understand what the Lenschief ray angle of your sensor is and what the CRA of the lens is If the CRA of the lens is smaller than that of the sensor, there will definitely be color deviation. Either change the lens. If an unsuitable lens is found in the market, it means that the quality of the sensor itself is not very good. 4. In theory, lens shaping is to solve the problem of different transparency of the lens. But perhaps each company can add their own algorithm and try it out. Like the angle of the incident light, lens manufacturers usually provide CRA curves because the CRA of the lens varies from center to periphery Besides Shading, it may still be necessary to adjust the AWB if the color is slightly red, because the bottom of the picture is actually a white area, and the sensor appears slightly red in the white area. Try adjusting the AWB again, or check if the three lines of R, G, and B overlap in the lightbox! If it’s an AWB issue, why is there still a white area in the image? AWB is not adjustable, some colors are biased, and some are not biased. Don’t say anything if you don’t know. If CRA is not matched, the color cast should be symmetrical, and if the bottom is red, the top will definitely be red. I personally think it should be caused by light leakage, either the barrel or the introduction of stray light through the light hole.

DOVDD28 wiring is too thin and too long, and the ground wire is unreasonable:
Phenomenon: Flower screen
Reason: The 2.8V voltage is due to the resistance on the wire absorbing the voltage, resulting in insufficient driving capability. The ground wire is pulled high and produces burrs, affecting signal integrity and data acquisition.

There is a voltage issue with the DVDD:
The highlight in the picture is the office window. All other parts are black without any details? What is the reason? AWB？ AGC？ Is it still the contrast? The problem has been solved, it’s the voltage of the DVDD is incorrect. The datasheet states 1.8V, but when asked about FAE, the result is 1.2V.

White stripe problem caused by small gain:
When facing a white object, the stripes shown in the following image will appear when entering the preview. When moving the children’s camera, these stripes disappear and will not appear in the future. They may only appear when entering the preview again. What is the reason for this? This issue has now been resolved by increasing the gain in the initialization code.

Image misalignment caused by frame rate issues:
Sensor 0v9655 may sometimes cause image misalignment when capturing 1.3 million images of sxga (as shown in the picture), while it does not occur with vga. Please help analyze. Thank you! The frame rate is too high and the exposure time is too short Can you adjust VBLANK and HBLANK to solve the problem of further reducing FPS to 5? Try it, what about your buffer speed?? Thank you all! Reducing the frame rate is more effective here.

Power noise:
Horizontal patterns appear. The problem has been solved. The power issue can be resolved by adding tantalum capacitors to AVDD. It is estimated to be caused by severe power ripple.

There are constantly changing fine horizontal stripes in the image:
Unlike the large area of rolling horizontal stripes caused by the flicker of fluorescent lamps, it presents a pixel high horizontal stripe like noise point, which is not fixed in position, has a relatively large number, and varies with the strength of the light.
Because setting certain sensor registers can affect the color of these horizontal stripes, it is basically ruled out that the interference caused by the board to the data during data transmission, and the possibility of poor contact is also ruled out. It should be that these horizontal stripes already exist inside the sensor. In addition, with the same initialization sequence and the same sensor, this situation did not occur in the manufacturer’s demo version, so it is basically ruled out that there is a software problem. Finally, it was found that in order to save hardware costs, the two analog and digital voltages of the sensor were output and supplied by the same chip, which caused mutual interference between the two and affected the normal operation of the sensor.
Solution: Separate analog and digital electricity for separate power supply

There are fixed jagged vertical stripes on the image:
There are obvious vertical stripes on the image, distributed in full screen, very fine, like blinds.
Upon closer inspection, it can be observed that the vertical stripe is actually a jagged stripe caused by the misalignment of adjacent pixels on the image. Upon closer analysis of the spec, it can be seen that since the sensor sends image data in bytes, in RGB565 mode, two bytes represent one pixel. In the Camera controller of the CPU I am using, data is processed in units of 4 bytes, or one word. Since the CPU processes data in LSB format, the order of two pixels within a word is reversed without adjustment. When DMA finally sends the data to the continuous buffer in memory, the order of pixels is: pixel 2, pixel 1, pixel 4, pixel 3…
Solution: Use a program to adjust the pixel order. To reduce the burden of additional calculations on the CPU, this step can be combined with other similar color conversion or PACK mode to Planner mode operations.

Image misalignment is prone to occur in large sizes:
When the sensor operates at its maximum resolution, the image is prone to vertical misalignment.
The tracking program can see that there is an overflow phenomenon in the FIFO cache of the CPU’s Camera controller at this time, which means that DMA cannot transfer the data in the FIFO to the memory in time. In this example, the clock of the sensor’s output data works at 24MHz at the maximum resolution. In theory, DMA should be able to transfer data urgently, but the memory bandwidth may also be occupied by other devices such as the CPU, causing it to not be able to write to the memory in time, resulting in DMA not being able to work at its maximum load, so it cannot read out the data in the FIFO in time, causing some data loss and image misalignment.
solution
In some cases, changing the startup threshold of DMA transfer can solve this problem, but in other cases it is ineffective. Considering that the highest resolution is only used for taking photos and not used for previewing, it is possible to change the resolution at the moment of taking photos while modifying the clock frequency of the sensor to a frequency that will not cause FIFO overflow, without affecting the number of preview frames. In addition, when intercepting the highest resolution image, try not to perform other memory related operations. Immediately switch back to the preview resolution after capturing the image. By using these methods, the possibility of FIFO overflow is reduced.

When the read data is displayed, it appears as a splash screen:
The data read appears as a blurred screen, but it clearly changes with the subject being photographed.
The displayed data is a complete flower screen, or the approximate outline of the object can be seen, but the color is completely incorrect, such as a patch of green. This situation is often due to mismatched image data formats, such as not processing YUV2RGB, and the sampling order of each component of YUV does not match the value order calculated by the software.
If the specific manifestation of a flower screen is that the image is constantly changing without regularity, it is usually possible that the triggering edge of the data reception is incorrect, resulting in incorrect data reception.
Another time, while scrolling, I carefully observed the pattern on the screen and noticed signs of partially misplaced and repeated patterns. Therefore, the analysis may be based on the physical layout of the sensor, whose aspect ratio is exactly opposite to that of the LCD. Upon careful examination of the Spec, it was confirmed.

Image interference issue:

The reason may be:
(1) There is a problem with the camera module of the children’s camera. Try replacing it with a new camera;
(2) The data cable driver capability is insufficient, which can be changed in the camera register. Ask the camera module FAE to see which areas need to be improved;
(3) When two children’s cameras share a data cable, the non working camera will weaken the working data signal;
(4) If the PCB wiring is too long, there may also be interference, but I think this possibility is small. After debugging, there is only one model like this, as the module manufacturer FAE said. However, their module itself also has problems. There are two factors involved: PCB (long wiring) and module sampling problems.

YUV order incorrect:
When the yuv order is incorrect, the following phenomenon occurs.

Take a look at the specifications of the children’s camera and change the values of the corresponding registers. The following are different yuv orders in the red box. Find them and change them accordingly.

Scattered light, ghosting:
The professional terminology of its industry is collectively referred to as Flare, which refers to the phenomenon of light and shadow appearing at the edges or the shadow of a complete object when shooting light sources or strong objects. This phenomenon can only be alleviated and cannot be completely eliminated. The reason is that the material of the lens causes not only refraction but also reflection of light, and the distance between the lens surface and the protective lens of the whole machine will be more obvious!

Using frequency mismatch:
The receiving sampling frequency must be configured correctly, otherwise the correct image cannot be obtained.

The above figure shows the phenomenon of mismatched sampling frequencies. Please note the difference between the phenomenon of incorrect YUV and RGB order configuration.

The following figure shows the reversed order:

Brightness and Night Mode:
I believe everyone now knows that the brightness of an image is related to the exposure time. Therefore, in order to display images in dark areas clearly, it is necessary to increase the sensor’s exposure time to dark images. That is, the line time will be set much larger than in normal mode, which can give the CMOS sensor more exposure time and improve brightness.
We have two methods to control image brightness, one is to use AE target, and the other is to increase grayscale gain. We use AE target method, which makes the colors of the images more realistic. Using this method will affect the frame rate. When the frame rate reaches our limit, we need to increase the analog gain, which will also amplify the image noise. Generally, there is no need to increase the digital amplification gain to adjust the brightness of the image.

If there is any infringement in the content section, please contact loris.lok@pardotech.com Delete