让高分辨率的相机芯片输出低分辨率的图片对于像素级的值有什么影响?

很多图像传感器可以输出多个分辨率的图像,如果选择低分辨率格式的图像输出,对于图像本身会有什么影响呢?

传感器本身还是使用全部像素区域进行感光,但是在像素数据输出时会进行所谓的降采样(down-sampling)。具体的降采样方式各个传感器可能略有不同,但是得到的效果都是另实际的PPI(pixels per inch)数目下降,即输出图片的实际像素数小于传感器最大像素数。

可以参考以下网址的一个回答

If I set my camera to lower resolution, does that mean more sensor area per image pixel?

In terms of PPI, no. In fact, your effective PPI in the final picture is reduced. When you choose smaller image sizes in camera, when the picture is taken, the full sensor is still used. However, when the camera's processor goes to write the image to a JPEG file, the image is first downsampled. This has the effect of being able to discern fewer fine details in the reduced picture.

Nikon's D3200 specifications page gives us the following (I added the sensor conversion to inches):

复制代码
Image sensor: 23.2 x 15.4 mm CMOS sensor (0.913 x 0.606 in)
Image size (pixels) [L]: 6,016 x 4,000 (24,064,000 pixels)
Image size (pixels) [M]: 4,512 x 3,000 (13,536,000 pixels)
Image size (pixels) [S]: 3,008 x 2,000 (6,016,000 pixels)

Thus, depending on whether L, M, or S is chosen in the camera settings, the D3200 will perform as either a 24 MP camera, 13.5 MP camera, or 6 MP camera. In all 3 cases, the entire CMOS sensor area is used to take the image. In other words, regardless of L/M/S setting, it is still a crop-sensor (1.5 crop factor) camera.

复制代码
For the Large setting, no downsampling is performed. Thus the effective PPI of the sensor image is: 4,000 pixels / 0.606 inches = 6,601 PPI.
For the Medium setting, a 4:3 downsampling is performed. Thus the effective PPI of the sensor image is: 3,000 pixels / 0.606 inches = 4,950 PPI.
For the Small setting, a 2:1 downsampling is performed. Thus the effective PPI of the sensor image is: 2,000 pixels / 0.606 inches = 3,300 PPI.

Returning to your question, but addressing "Sensor Pixel Area" (I read that as effective pixel (sensel) size),

复制代码
Would setting it to 13.1MP allow for a greater Sensor Pixel Area ... in my final picture?

Yes. At the following image sizes, your effective sensel size is approximately:

复制代码
Large setting: 14.8 µm²
Medium setting: 26.4 µm²
Small setting: 59.3 µm²

The area covered by each pixel in the final image is the size of several pixels on the sensor. Fewer pixels in the same area means there is more area covered per pixel. Or to put it simply, if you reduce PPI, then you increase the area covered by each pixel. Pixel/inch inverted is inch/pixel. 100 ppi means the same thing as 1 inch per 100 pixels which means the same thing as .01 inches per pixel. 50 ppi means the same thing as 1 inch per 50 pixels which means the same thing as .02 inches per pixel.

相关推荐
老黄编程7 小时前
大型工地实时数据处理与三维重构系统方案(极简中心化部署版)
人工智能·数码相机·计算机视觉·大数据处理·入侵检测·三维重构
Godspeed Zhao9 小时前
具身智能中的传感器技术40.2——事件相机0.2
人工智能·科技·数码相机·机器学习·事件相机
一起搞IT吧1 天前
相机Camera日志实例分析之二十:相机Camx【照片后置4800/5000/6400万拍照】单帧流程日志详解
android·嵌入式硬件·数码相机·智能手机
格林威1 天前
3D相机视觉检测:环境光太强,结构光点云全是噪点怎么办?
开发语言·人工智能·数码相机·计算机视觉·3d·视觉检测·工业相机
格林威1 天前
线阵工业相机:线阵图像出现“波浪纹”,是机械振动还是编码器问题?
开发语言·人工智能·数码相机·计算机视觉·视觉检测·工业相机·线阵相机
格林威1 天前
工业视觉检测:单样本学习 vs 传统监督学习
人工智能·深度学习·数码相机·学习·计算机视觉·视觉检测·工业相机
格林威1 天前
工业视觉检测:两大主流异常检测开源框架深度对比(PatchCore vs SPADE)
开发语言·人工智能·深度学习·数码相机·计算机视觉·视觉检测·工业相机
格林威1 天前
线阵工业相机:如何计算线阵相机的行频(Line Rate)?公式+实例
开发语言·人工智能·数码相机·算法·计算机视觉·工业相机·线阵相机
格林威2 天前
工业视觉项目:如何与客户有效沟通验收标准?
人工智能·数码相机·计算机视觉·视觉检测·机器视觉·工业相机·视觉项目
科研前沿2 天前
什么是时空融合技术?
大数据·人工智能·数码相机·算法·重构·空间计算