Without an image for comparison, "blurry" can refer to "out of focus" for instance because the distance between object and lens is below the current mode's working distance. For point and shoot cameras like the one you link to, this threshold differs between for recording photos in the standard / automatic mode vs. the one about macro photography. In addition to this, it depends on the setup of the lens (wide angle / short focal length vs narrow angle / long focal length). If you underpass this threshold, the camera will not yield a sharp image. Indeed, some models block the trigger until the camera's focus is set.
A second plausible cause is the object. If it is dimly lit and the camera's algorithm to focus is based on visible light passing the lens, it likely will fail. The algorithm equally will fail if the sensor area to focus only is a blob in the center of the visual field (sometimes useful for a portrait), but your object is very even. For instance if your page is mostly empty because there only is a final paragraph on top of said page. Check the camera's manual if there is a setup which considers multiple spots in the field of vision to set the focus. For comparison:
(image credit Wikipedia)
A third plausible contribution for "blurry pictures" is the camera's sensor itself, by systematic and random noise all the way between capturing the signal, processing, and storage of the image. How well the object is lit and time of exposure to record a photo already were mentioned as an influence. The dimension of the individual pixels on the chip, the design of the color filter array, the total number of pixels at disposition, the number of pixels per unit of surface; the selected sensitivity (ISO number) and time of exposure per image are additional parameters of the first step. Depending on the camera's software, you can bin multiple pixels to obtain an optical (spatial) resolution more coarse; check if recording with e.g. 20MP instead of 48MP yields a better image (depending on the dimension of the page, it might suffice -- see Daniel Reetz' lessons learnt on the archivist project.) Sharpness usually is lost in the second step once the readout of individual pixels is binned, merged, compressed to yield a jpg file often faster to transfer/arguably more commonly processed by (simpler) image programs than a for instance tiff set up as raster image.*
To some degree, you can counter the influence of the third contribution by statistics; record the photo of your object multiple times, then compute an average photo eventually retained to increase the signal/noise ratio. If you have e.g., five images about the same page of identical dimension and scale in a dedicated sub folder, imagemagick's convert
command allows you to run
convert *.jpg -evaluate-sequence median output.jpg
to briefly superimpose them. The same spot (as in address of a pixel) in input image 1, input image 2, input image 3, etc is read out to compute the median eventually written into the same address (i.e. pixel) of the new image output.jpg
.** Why 5 input data? In a series of N random experiments, the signal/noise ratio scales by the square root of N, i.e. by four times as many measurements about the same object, this ratio increases by 2, and an additional 5th recording of input data is an affordable safety margin.
Similar to the above, you equally could issue the command
convert *.jpg -evaluate-sequence mean output.jpg
where imagemagick now computes the arithmetical mean value (or average) for the new pixels of output.jpg
. However this is more reasonable if you have many images (like thousands in speckle imaging) instead of only a couple aiming for a book scan. Here, the computation of the median is less influenced by outliers processing input image 1, 2, 3, etc than the computation of the arithmetical mean value.
* A tiff file can be seen as a container. Its data can be stored uncompressed as read-out from a detector/chip. To save storage memory per file, a tiff optionally can be compressed either losslessly (e.g., by LZW algorithm), or lossy (e.g., jpg). If interested, Nico Stuurman's lecture Cameras and Detectors I: How Do They Work? and Cameras and Detectors II: Specifications and Performance shares some information about image recording and noise.
** This technique equally allows you to "clear a street" of randomly moving pedestrians and cars based on a couple of photos recorded with a camera mounted on tripod.